Click here to load reader
Jan 12, 2015
STORAGE AND WIRELESS APPLICATIONS
By Kishore Jethanandani
In collaboration with Datacomm Research Company
TABLE OF CONTENTS
1 INTERDEPENDENCE OF WIRELESS AND STORAGE..................................... 6
1.1 Value of Storage--Aggregation....................................................................................... 7
1.2 Value of Storage—Assimilation...................................................................................... 8
1.3 Value of Storage—Dissemination .................................................................................. 9
1.4 Storage and Wireless Applications .............................................................................. 10
1.5 Managing Storage Area Networks ............................................................................... 11
2 TECHNOLOGY ....................................................................................................... 12
2.1 Beyond SCSI................................................................................................................ 13
2.2 Networks and Scalability .............................................................................................. 14
3 STORAGE AND WIRELESS APPLICATIONS.................................................... 19
3.1 Storage and Digital Photography ................................................................................. 19
3.2 Unified Messaging, Voice Information and Vehicle Telematics ................................... 21
3.2.1 Storage and Unified Messaging .......................................................................... 23
3.2.2 Storage and Voice Information services.............................................................. 25
3.2.3 Storage and Telematics....................................................................................... 26
3.3 Location based services............................................................................................... 26
3.4 CUSTOMER RELATIONSHIP MANAGEMENT .......................................................... 28
3.5 STORAGE AND MEDICAL APPLICATIONS............................................................... 30
3.6 Rich Media Applications............................................................................................... 34
4 PROSPECTS OF LEADING APPLICATIONS OF STORAGE .......................... 36
4.1 Prospects of Digital Photography................................................................................. 38
4.2 Unified Messaging, Voice Information and Telematics ................................................ 40
4.3 Business Intelligence ................................................................................................... 45
4.4 Prospects of Location Based Services ........................................................................ 46
4.5 Medical Applications .................................................................................................... 48
5 COMPETITIVE ISSUES ........................................................................................ 50
5.1 Digital Photography...................................................................................................... 50
5.1.1 Applied Science Fiction’s (ASF™) ....................................................................... 51
5.1.2 Pixel Magic Imaging (www.pmimaging.com)....................................................... 53
5.2 Unified Messaging, Voice Information and Telematics ................................................ 54
5.2.1 Tornado Development (www.tornadodevelopment.com/)................................... 54
5.2.2 BeVOCAL (http://www.bevocal.com/index.html) ................................................. 57
5.3 CRM/Business Intelligence .......................................................................................... 58
5.3.1 Teradata (www.teradata.com)............................................................................. 61
5.4 Location Based Services.............................................................................................. 63
5.4.1 Space Machine (http://www.spacemachine.net/) ................................................ 64
5.5 Medical Applications .................................................................................................... 67
5.5.1 General Electric Medical Information Systems
http://www.gemedicalsystems.com/it_solutions/index.html)................................................. 68
6 EXPERIENCES OF KEY PLAYERS: ................................................................... 72
6.1 Storability (www.storability) .......................................................................................... 72
6.2 IBM/Tivoli...................................................................................................................... 73
6.3 Veritas (http://www.veritas.com/) ................................................................................ 77
6.4 E-motion (www.emotion.com) ...................................................................................... 78
6.5 Viafone (www.viafone.com) ......................................................................................... 79
6.6 Tellme (www.tellme.com)............................................................................................. 81
6.7 Openwave (http://www.openwave.com) ...................................................................... 82
6.8 Datadirectnet ( http://www.datadirectnet.com/)........................................................... 83
6.9 Generic Media (www.genericmedia.com) .................................................................... 84
6.10 Digital Fountain (http://www.digitalfountain.com)......................................................... 85
6.11 Airborne Entertainment (http://www.airborne-e.com)................................................... 86
6.12 I-DRIVE (http://www.idrive.com/) ................................................................................. 87
6.13 Deep Bridge (http://www.deepbridge.com/) ................................................................. 88
6.14 Broadstream (http://www.broadstream.com/) .............................................................. 89
6.15 Vicinity (www.vicinity.com) ........................................................................................... 90
6.16 Mediabin (www.MediaBin.com).................................................................................... 92
6.17 Scale Eight (www.scale8.com)..................................................................................... 94
7 CONCLUSIONS: IMPLICATIONS FOR BUSINESS STRATEGY .................... 95
EXECUTIVE SUMMARY
Storage technologies will drive the adoption of wireless applications in the near term future.
The value proposition of storage technologies is the aggregation, assimilation and
dissemination of large volumes of information. Wireless technologies help to feed data to
large repositories and are essential for real time communication of messages for speedy
execution of tasks.
The impact of storage technologies on wireless applications will be more significant in the
future as the transition from LAN attached storage devices to storage area networks (SAN)
is completed. SANs, aided by related technologies, are an efficient means for data delivery
and its conversion, from a single source, for reuse by a large variety of wireless devices.
Innovations in system integration as well as storage management software, virtualization
and content management tools, will play a key role in speedy and cost-effective delivery of
data to wireless devices. File systems will enable rapid retrieval of data from tapes for cost-
effective use of imagery, inter-operability in a heterogeneous environment and intelligent
caching to overcome speed limitations of disks.
The six most important storage enabled wireless applications are customer relationship
management/business intelligence, digital photography, enterprise location based services,
unified messaging, voice information and vehicle telematics as a single group, medical
applications and rich media.
Storage requirements for digital photography will increase rapidly as the marriage of film
and digital photography, ubiquitous kiosks and digital photography networks increases the
rate of adoption of digital photography. Digital photography networks will facilitate mobile
albums that can be readily shared.
Storage helps to reap economies of scope from the convergence of multi-media
messaging, mobile voice applications and vehicle telematics. The current high costs of
unified messaging will be lowered by storing all messages in a single undivided data
repository. The increasing popularity of infotainment by vehicle owners, beginning with
satellite entertainment, will increase the demand for storage.
The size and nature of CRM databases has been transformed for real time decision
support. Increasingly, terabyte size repositories process a variety of data at shorter
intervals. Data processing is required for not only strategic purposes but also for tactical
reasons. Data repositories need wireless technologies to not only to receive data quickly
but also to alert decision makers in real time.
Enterprise location based services, not consumer services, will play a key role in increasing
the demand for storage services by the wireless industry. The increasing accumulation of
attribute data correlated with spatial data, imaging, raster data and satellite imagery and
aerial photography will increase the demand for storage. Furthermore, storage helps to
streamline data flows for use in low bandwidth wireless devices.
Medical applications are poised to move beyond imaging data to integrating clinical
information systems and monitoring data for business process efficiency. The deployment
of wireless applications will increase rapidly as they are used for decision support. Storage
demand will increase rapidly as the variety of data stored increases.
Rich media applications of storage will grow mostly outside the broadcasting industry for
Internet broadcasting and enterprise applications. This market has a better longer-term
prospect than in the near term as system integration issues are mired in politics.
1 INTERDEPENDENCE OF WIRELESS AND STORAGE
Introduction: Internet storage technologies, shorn of their technological mystique, are a new age
version of a library. They share the attribute of a repository, which aggregates and preserves
records of information. The analogy does not go one whit further. Storage systems aggregate
information on a scale unimaginable with libraries and they preserve records over a much longer
period of time. Movies or even images can be stored for an indefinite period of time, which is
inconceivable with traditional libraries1.
1 A case study of a digital archive can be found at http://www.dvwebvideo.com/2000/0500/gordon0500.html
Unlike traditional libraries, the size of storage area networks is not limited by space. It expands
with not only the memory capacity of disks but also with the speed of networks that interconnect
arrays of disks as well as software that organizes the data logically for its management from a
single point.
Information on a storage area can also be preserved for a longer period of time since it is
replicated on several disks within or outside a region. If the World Trade Center were a traditional
library of America’s heritage documents, it would have been completely destroyed. On a storage
area network, the same documentation would have been available within the hour at another
destination.
1.1 Value of Storage--Aggregation
Data storehouses are fragmented by the publishing medium, its location or by the type of
equipment or the software used to manage content. Storage technologies aggregate information
from diverse sources.
Data is available on printed records, on films, images or databases. In the world of a traditional
library, stored documents are typically printed documents or at best databases. Films and images
are rarely available and harder to juxtapose with other sources of information. Storage area
networks digitize all information and permit their rendering on a single document.
Information is also fragmented when it’s generated or collected in separate regions. City or state
governments typically gather geographical information and don’t necessarily want to share it at a
centralized point. Storage area networks either provide pointers (by providing metadata or data
about data) about sources of information or segment a unified network such that secure zones
are accessible to those who have acquired the rights to do so.
Free flow of information across regions pre-supposes high speed networks before traffic can flow
efficiently across regions. Emerging storage companies are lighting up ample dark fiber networks,
GiantLoop Networks (www.giantloop.com) has launched its products, to move large volume
data such as that generated in the health industry. Alternatively, storage area networks can mirror
information on disks across geographical boundaries so that information can be accessed locally
without the delays of data flow from the core.
Silos are also created by the technological limitations of equipment. Servers combine both the
storage and the processing function. Consequently, information stored on them does not have to
flow out. Typically, servers are connected to LANs, which have low bandwidth and cannot
manage the flow of large volumes of data. Documents on a storage device can only flow out as
only servers can use it and their networks are designed for large volume data flows.
1.2 Value of Storage—Assimilation
Growing volumes of data don’t go awry because storage volume management, virtualization and
content management software organizes them. Volume management software is equivalent to
the floor plan and the scheme for placement of shelves in a traditional library. It automatically
allocates data to disks without manual intervention. Storage virtualization software is like the
indexing system that directs users to the location of content in storage area networks. Content
Management software is the Dewey Decimal system, which identifies the content available in a
library.
Storage virtualization software, developed by companies such as Veritas (www.veritas.com)
encapsulates terabytes or petabytes of data into a directory structure analogous to the explorer
on the desktop. It lays out the logical structure of data for the corresponding physical location
much like URLs correspond to IP addresses on the Internet. Typically, the user interface of
virtualization software is a portal type of software with a web browser. Unlike an index in a
traditional library, portal software not only simplifies searches of information but also the
movement of data from the point of storage to its consumption destination. Data can simply be
moved by drag and drop methods (security considerations permitting).
Large repositories would risk loss of the value of their information without speedy retrieval of
information of diverse types. Content management defines each data type and its characteristics.
Metadata or the data about data identifies the traits of the data stored such as dates, location
theme so that it can be retrieved by using the attributes as the keyword.
Content Management software automates the processes of archiving, indexing, searching and
assimilation of information. Archiving is done on pre-formatted templates which have meta-data
attached to them. Indexing is automated by combing documents for the recurring words in the
document to identify the themes in the document. Search processes are eased by the
classification of data that is enabled by metadata. Finally, metadata can also define access right
and intellectual property rights associated with content.
The value of content depends on the speed of retrieval as well as the ability to pick specific
pieces of information. General Motors, for example, increased the revenue from its content from $
4 million to $25 million by using Artesia’s content management tools. When indexing is done at a
more granular level, such as by video logging technologies available from Virage, reuse of
content is possible, which enhances its utility as an archive.
Content Management tools are also a means to assimilate information. Typically, information is
pieced together by identifying the common denominators in the entire data. Thus patient social
security number is a way to identify all related health information. Similarly, spatial co-ordinates
help to organize all geographical information.
When a diverse range of information is collated at one point and can be cross-referenced with all
related information, its assimilation and processing becomes possible. Text, numbers and
pictures can be put juxtaposed to bring into relief unnoticed relationships. Data can be pieced
together to uncover the big picture and to identify patterns. Similarly, statistical software can be
used to find correlations in the data. Storage technologies can help do this on the fly. Fraud
detection is one application that is enabled by cross-referencing of data available in a single
repository.
1.3 Value of Storage—Dissemination
Finally, dissemination of information is more convenient, faster and has a wider reach with
storage technologies. This is particularly true with storage area networks; several copies of
information can be mirrored at several sites and with additional aids can be converted into several
formats. Traditional mediums like analog TV are unable to offer the same content in any other
format. Digital content, on the other hand, can be adapted for its presentation of several wireless
devices or other devices2.
Museums, rare books and historical archives are striking illustration of illiquid information. Today,
the global mirroring feature of storage technologies, a service offered by Scale Eight among
others, has enabled not only their centralized storage but also their worldwide dissemination
(www.scale8.com/customers/Octavo_Case_Study.pdf).
The applications of storage are rooted in these generic attributes of archiving, retrieval,
assimilation, and dissemination that have been enhanced by digitization and networks that
interconnect storage devices. Ease of archiving, for example, encourages the preservation of old
movies that would be lost by deterioration of silver halide films. The cost of maintaining large
archives would not be bearable if footage was not conveniently retrieved such as by natural
language keywords3. Similarly, geographic data and related traffic information would be hard to
use for logistics management if could not be rapidly aggregated, assimilated and disseminated to
truck drivers.
1.4 Storage and Wireless Applications
Wireless Internet users have potentially a greater need for gathering, archiving, assimilating,
retrieving and speedy dissemination of information. Sources of information are more scattered
when data is gathered from mobile professionals and needs to be aggregated to be of use by the
entire enterprise. Field forces will be unable to plan their schedules unless all data of all pending
orders is available at one point.
Rapid assimilation is essential since mobile staff often execute tasks with short lead times. Tasks,
such as ambulatory healthcare, are most efficient when staff on board is alerted about a patient’s
past conditions as the event happens. Similarly, mobile sales staffs require information on
demand and inventory condition in real time to serve orders.
2 See the case study of Generic Media for the method for serving data for a variety of players in the streaming media context. 3 See the case study on E-Motion
Just as important is the rapid retrieval of information for mobile staff. A typical instance is the
need to provide documentation in the sale of pharmaceuticals. Mobile sales staff doesn’t carry all
the documents; it would rather remotely access them from a storage device and print them on a
customer site4.
Finally, dissemination of information to wireless users implies that content has to be distributed to
several different types of devices. When content is available at a single point, its format can be
converted to suit the needs of each device.
1.5 Managing Storage Area Networks
The very size of storage systems throws up entirely new challenges for managing information.
Storage management software is required to spread the load across the numerous disks, tapes
and optical disks across a storage area network to minimize time delays and costs. Variants of
this software manage replication of data in geographically dispersed sites, back up of data to
ensure its availability and recovery when the data is lost. All this has to be achieved on platforms
as diverse as Windows NT and UNIX. The liquidity of information depends on the Application
Program Interfaces (APIs) that help to interconnection with the broad variety of operating systems
that are deployed on storage area networks. This is very largely an unfinished task in the
management of storage area networks.
For storage systems to be more than archives or tape systems, information has to flow from disks
to applications in a predictable and speedy manner to be useful to applications. Storage
management software ensures that information flows to applications in an efficient manner. Tivoli,
for example, develops software to move data from storage disks to applications quickly.
Other types of software manage the quality of service on a network are also required for industrial
scale application of storage area networks. These are required to monitor and report on
performance or the extent to which service level agreements have been met5.
Content Management software is slow to grow because data definitions are hard to standardize.
Individual companies have developed content management software for silos and have a variety
4 See our case study on ViaFone 5 See our case studies of Storability and Broadstream
of data definitions embedded in their legacy systems. In a shared environment like a storage area
network, the data definitions have to be understood by several different types of users.
Fortunately, XML enables the sharing of data independent of the particular representation of data.
Internet storage technologies need the efficiencies brought about by storage management
software, management software and content management software to drive the adoption of
applications such as especially Internet broadcasting. The pace of development of storage
software has been relatively slow and has set the limits to diffusion of storage-enabled
applications.
Storage technologies also need high-speed networks to communicate over longer distances.
Hardware implemented communications systems like Fiber channel have distance limitations.
Other software-implemented protocols such as TCP/IP are much slower since they correct for
errors. The dissemination of information over longer distances will be constrained till that
happens.
2 TECHNOLOGY
The key to the ability to aggregate information in colossal repositories is increasing efficiency of
networks. Data would have to be stored locally, within or near a server, if networks are not
efficient enough for retrieval from a remote location. In the early stages of information
management, data was stored in the memory of the server.
Storage within the memory of a server strains the processing power of a server for larger volumes
of data. Consequently, rising volumes of traffic impair the ability of servers to operate applications
efficiently. Also, servers have more than one source of failure when they combine the functions of
operating an application and storage.
The combined efficiency of servers and storage devices is increased and their downtime is
lowered by specialization; the functions of running an operating system (and user applications)
are separated from that of the management of files. Storage becomes the function of a
specialized device separated from a server.
The division of functions can also be between multiple storage devices and multiple servers
working together. A switch routs the traffic when a cluster of storage devices works with a group
of servers.
2.1 Beyond SCSI
The early development of storage area networks took place with parallel SCSI (Small Computer
System Interface) networking technology. This technology has bus architecture; a single server is
electrically connected to a corresponding storage device. Since servers in the SCSI world work
jointly with only a single storage device, any loss of capacity necessarily involves downtime. The
data transfer rates cannot be any more than the speed of disks because they cannot retrieve data
from any more than a single storage device.
SCSI has other limitations like the number of addresses and distances that it can cover. In a 16-
bit environment, it can have no more than 15 devices. The number of devices on the network
limits its scale of operations. SCSI also cannot operate at more than 25 meters of distance. SCSI,
therefore, does not effectively disseminate information.
Fiber channel networks overcome the limitations of SCSI type of networks. They are capable of
switching (besides a ring technology) so that a cluster of servers works with a corresponding
array of storage devices. Since each port on the switch can be connected to another switch, the
number of addresses can be increased without any limit. Inter-switch connections, however,
come at the cost of a latency penalty that is incurred as traffic flows from one port to another. In
addition, failure on any one port has a domino effect on all interlinked ports.
Therefore, the building blocks of larger scale storage area networks are directors or switches with
more than 32 ports. Directors are indispensable for applications such as rich media, which require
parallel retrieval of data to reduce the latencies involved in recovering data from disks.
Companies such as McData (www.mcdata.com) specialize in the design and production of such
switches.
Another major difference between fiber channel networks and the SCSI type of topology is that
the former inter-connects servers and devices by a network and data transfers don’t take place by
electrical signals. Consequently, devices can be added, as demand grows, to the network without
interrupting operations just as appliances are connected in a grid.
Finally, the devices connected to a Fiber Channel Network can be spread over longer distances
over as much as 10,000 meters. It is possible to offer disaster management services on such
networks as data can be replicated on several devices such that they are dispersed
geographically.
The hardware capabilities of fiber channel networks have to be complimented by software
management tools before services such as disaster management or replication of data can even
be offered. Storage virtualization software helps to achieve these functions.
Storage area networks (SANS) have had less success than its precursor, the Network Attached
Storage (NAS) systems, in developing the file systems required to manage large networks. Some
companies such as StorageNetworks (www.storagenetworks.com) have implemented file
systems to manage storage area networks for large enterprises from behind the firewall. Efforts to
manage an open storage area network have been initiated by Hitachi Data Systems
(www.hds.com) and IBM.
2.2 Networks and Scalability
In the technical literature, more so in the technical marketing literature, facetious distinctions
between SANS and NAS have been drawn based solely on the features of the technology, SANs
are seen to move blocks of data while NAS manages a file system. The fact is that a NAS server
manages a file system, which in turn maps blocks of data and helps to manage their flow.
Conversely, files systems or storage virtualization software manage the flow of blocks of data on
a SAN. If there is any distinction between the two systems, this is a difference between
tweedledee and tweedledum!
Similarly, SANS and NAS are distinguished based on the networking technology that
interconnects the devices; fiber optics are associated with Fiber Channel and Ethernet with NAS.
The reality is that Fiber Channel, implemented as it is in hardware, has a lower latency rate while
Gigabyte Ethernet costs less but has a higher latency rate. Storage area networks, with their
larger volumes, have a lower tolerance for latency. On the other hand, NAS processes lower
volumes and can make-do with relatively higher latency rates. There is nothing to prevent SANs
from using Ethernet networks and vice versa for NAS.
The essential distinction between SAN and NAS is that the former has a separate network for
storage devices while NAS devices are attached to an existing LAN. It is this property of SANs
that enables it to scale and to provide other services such as continuity, disaster recovery and
load management.
SANs can potentially operate at a much larger scale of operation than NAS because devices can
be added to a network without disrupting ongoing operations. The management of larger volumes
of data would not be possible without the virtualization software that helps to manage the
allocation of data between devices and the movement of data between them on storage area
networks.
A Network Attached Storage (NAS), a stripped down variant of a file server, manages data flow
from storage devices independent of an application server. The storage device appears as an
additional drive in the directory on the operating system of an application. Redirector software
manages the data flow from the NAS to the client. As storage devices are added, the software
directing I/O calls from the client have to be adjusted to manage disk space, which involves some
downtime unless there is no overlap in the data stored in the storage device added and the
previous one.
On the other hand, a Storage Area Network is intended to expand seamlessly without incurring
downtime as capacity is increased. The file system resides on a server and directly allocates
blocks of data on disk space. Any addition to capacity is managed by a volume manager, which
allocates data to storage capacity. As SANs evolve further, the file management function is
undertaken by in-band or out-bands appliances specially meant for the management of the
network.
The slow progress in developing virtualization software for SANS accounts for its lower rate of
adoption compared to NAS which works with proven file systems. Operating systems for SANs,
especially in heterogeneous environments, are currently under negotiation and development.
In the NAS environment, network protocols for joining storage devices with servers on a network,
the Network File System (NFS) and the Common Internet File System (CIFS), provide a means to
share files with a variety of computing environments and to ensure security when transactions
take place on the network. The NFS6 for example, automatically mounts the servers file system
onto the client where it seems like a local directory. By requiring a client to have an account with
the NAS server ensures security. Before granting access to a server, the client’s identity is
checked.
By contrast, the progress in the development of standards for interoperability in Storage Area
Networks is much slower7. In 2001, vendors like Hitachi Data Systems, IBM and Brocade took the
initiative to plug the holes in fiber channel SAN systems. The key issues are fiber channel’s
vulnerability to security breaches. In addition, SAN systems currently manage networks by out-of-
band monitoring systems using the Simple Network Management Protocol.
SANs can play a vital role in lowering the failure rate in running applications since traffic can be
spread over several disks. Typically, the building block of a SAN or a NAS is a RAID (Redundant
Array of Independent Disks, a RAID controller manages traffic locally) or a JBOD (Just a Bunch of
Disks without any intelligence) or a collection of disks which don’t have local intelligence and all
data management is done by the file system. RAID stores blocks of data, which are divided into
smaller units called stripes of 512 bytes. The option to spread traffic over several disks helps in
load balancing and redirecting traffic when any one of them fails.
Spikes in traffic, common with e-commerce or rich media applications, are better managed when
data flows are spread over several disks and balanced centrally by software residing on the RAID
controller or in the file system. Capacity additions can take place incrementally in step with
demand because they don’t have to be lumped with server investments. Installations of additional
disks are not disruptive since they can be plugged into a group of operating disks much like
appliances are to a utility network. Although some disks are redundant, capacity utilization is
better than with server-attached storage due to the benefits of load balancing.
6 (http://uwsg.ucs.indiana.edu/usail/network/nfs/overview.html) 7 This is discussed in detail in a white paper at (www.brocade.com/san/white_papers/pdf/EvolvingSANStandards03072000.pdf).
Although SANs are meant to be a means to manage larger networks than NAS, the reality is that
the adoption rate of the latter is much higher. One reason for this is that protocols for integrating
networks with devices in a NAS are proven while corresponding protocols for integrating SANs
are currently under negotiation and development.
In the absence of interoperability, the benefits of aggregation of information sources such as
centralized management are built on proprietary standards. Consequently, applications
deployment is hamstrung by the inability to assimilate information from diverse sources.
SANS facilitate dissemination of information by their ability to mirror data on geographically
dispersed storage devices. The information can then be viewed simultaneously by an audience
such as the global employees of a company listening to their CEO. Internet broadcasting
becomes technically viable with spatially distributed storage capacity
SANs also play a vital role in the preservation of information by placing them at a number of sites.
By replicating information at geographically dispersed regions, any loss of information that can
take place by natural or technical disasters is undone by recovering information from another site.
Communication networks determine the geographical reach of storage area networks while
storage management tools affect the latency, speed of recovery and other services that
consumers can have. Lower latencies are crucial, for example, for delivery of rich media
applications.
The value of stacks of information is high when it can be retrieved with low rates of latency. The
efficiency of retrieval of information is limited by the speeds at which disks operate. IBM, for
example, admitted that disk speeds are not increasing at rates comparable to other components
of the system such as network speeds and processor MIPs. In the near term future, IBM foresees
disk speeds rising to 15K-25K RPM, incorporated in its Shark products, but probably no more.
Consequently, companies are looking to parallel processing of information which involves routing
of data through several ports working simultaneously (as profiled in our case study of
DataDirectNet).
When stored on disks, data is saved in blocks of information, which are subdivided as stripes
when they are written on disks. Before the stored data can be transmitted, it is reassembled as
blocks before it reaches a network node. Inevitably, the process of recalling blocks of information
and reassembling them involve mechanical delays.
Storage network management companies are overcoming these problems with metadata or
mapping information that reduces the time delays in identifying blocks of information and
reassembling them.
Networks can be clogged when numerous streams of data are accessed simultaneously. Such
an eventuality is very likely when a very high number of customers are drawn to the same
information, as was the case when Americans wanted to read Kenneth Star’s report on the
Internet. Storage management companies are learning to create master files that can be cached
at a central point before customers’ access it (as profiled in our case study on Digital Fountain).
Additional inefficiencies are incurred when content has to be presented to several different media
players at the customer end. Content has to be transmitted in a way that is appropriate for each
of these players which taxes bandwidth capacity. New technologies are emerging that create
master copies that convert the content for each player on the fly (see our case study on Generic
Media).
When data is transmitted for wireless applications, it has to be transmitted to several towers
where they are close to the users of mobile devices. If this is done serially, the delays will be
enormous. Storage management companies are finding ways to transmit data in parallel streams.
In the future, the battleground in the storage industry will be storage management software. The
design of file systems can help to increase the productivity of already installed storage systems.
This can be achieved, for example, by varying the size of stripes depending on the nature of the
traffic, by caching in real time so that data does not necessarily have to come from the disk. Other
possibilities include the management of the metadata. If the metadata is also read from the disk,
the disk reader moves back and forth from the file data to the metadata. Some companies have
incorporated the metadata into the file system so that data is accessed directly. Yet another way
is to organize related data contiguously so that it can be retrieved quickly. File systems can also
be designed to speed up retrieval from tapes by keeping their metadata in the file system8.
8 More information of innovations from emerging companies can be found at
3 STORAGE AND WIRELESS APPLICATIONS
Storage sub-systems, per se, provides services such as back up, archiving, recovery, replication
and mirroring which do not have a direct bearing for applications. However, storage sub-systems,
together with storage management software, are an aid to information management. We will
discuss, in individual cases, the specific impact storage has on applications
3.1 Storage and Digital Photography
Consumer fulfillment with digital photography is intertwined with storage technologies. The
attributes that consumers value in photography include besides the quality of photographs, ease
of editing, the ability to share them with family and friends as extensively as possible and with the
least effort, preserve them and to retrieve them when required. These services are offered by
companies such as Shutterfly (www.shutterfly.com/index.jsp), Kodak’s
(www.ofoto.com/Welcome.jsp) and Pixel Magic Imaging (www.pmimaging.com).
Storage needs increase as the quality of digital photographs improves with higher resolution and
superior color texture. CD-ROMs or other removable media have met the needs of early adopters
but the demand for Internet storage will increase at higher levels of usage and as photographs
are transmitted over networks.
Image size of digital photographs, in terms of data, is large even when compression techniques
are utilized. This is because light, after it filters through a lens, is recorded by photosensitive
pixels in shades of black and white. The resolution improves with the increase in the numbers of
pixels.
Furthermore, colors are superimposed on the black and white image. A combination of green, red
and blue filters (twice as many green filters as red and blue) intercede the light falling on pixels;
computers then estimate the actual texture of the color from the data on hues of all the
neighboring filters. The entire process is data intensive and requires storage.
The equivalent of the quality achieved by silver-halide film requires a resolution of 2.1 million
pixels9, which is about 2MB for each image. Insertion of colors in the right proportion further adds
to the file size increasing it to 6 MB. The storage capacity of most personal computers will begin
to run out after a few rolls are shot. Some of the burden of storage is relieved by compression. A
standard such as JPEG can achieve compression rates of 20:1 efficiently but photographs begin
to lose their quality beyond that stage. Even so, each roll of film would require about 6 MB of
space, which is still very high.
Currently, users have the ability to use flash media with their digital cameras. However, the flash
media has a limited capacity and a very expensive medium to store data. The data from flash
media has to be inevitably transferred to a computer or other storage media. Consequently,
storage is required especially as volumes increase.
Once digital photographs have been stored, they can be conveniently edited with tools such as
Adobe Photoshop. On the other hand, editing of traditional photographs requires sophisticated
dark room techniques before changes can be made on them if at all.
People cherish photographs because they can share glimpses of their lives with friends and
families and preserve them as mementos for themselves. Storage enables sharing conveniently
when it is stored on the Internet. More consumers share their photographs as they take recourse
to the Internet, e-mail and disks to store their images. By the year 2001, 92.8% of digital still
camera owners shared their pictures compared to 77.7% in 199910
Perhaps, the greatest advantage of digital photography used in conjunction with storage is the
ability to use metadata or the data about identification of photographs. Metadata keeps a record
of when, where and by whom the photograph was taken, the location, subject and other
information that can help to retrieve a photograph. In industries like law and regulation, the
preservation of visual records for long periods is critical and hard to achieve because films
deteriorate and are hard to retrieve11.
9 Quoted from http://www.webtechniques.com/archives/1998/09/wang/ 10 International Imaging Industry Association, Fact Sheet. 11 For more information on the use of metadata in digital photography, see “The Power of Metadata Is Propelling Digital Imaging Beyond the Limitations of Conventional Photography”
Wireless photography and storage have a symbiotic relationship in the enterprise space. It helps
in reducing the tedium of documentation in the insurance industry and the real estate industry.
Companies such as Flashpoint (http://www.flashpoint.com/home.html) provide digital
photography solutions, in partnership with Sprint, for transfer of images over telecom network and
are stored. A typical case is the use of wirelessly transmitted photography for faster damage
assessment in insurance claims. Similarly, digital photographs are used in the real estate industry
to provide a glimpse of properties that customers can evaluate.
In the consumer space, storage can potentially created a mobile album, i.e., resident on a
network which can be shared impromptu with friends and family. However, this would have to
await a ubiquitous network of the kind the International Imaging Industry Association is working
on (http://www.i3a.org/pr_11_13_01.html).
Storage and sharing of photographs involves trade-offs that have not been satisfactory for the
consumers. Two forms of compression have been commonly used with JPEG and they are
lossless and lossy compression. Lossless compression simply means that data is not lost as a
result of compression so that economy in storage is considerably less than with lossy
compression. However, lossy compression is at the risk of loss of data that may not, at the outset,
seem essential and will not cause visible loss of quality. For example, data representing blue sky
in a picture would be identical and saves storage if it is trimmed. The algorithms that make the
adjustments may not, however, be understood by another program and over time important
information will be lost permanently. JPEG 2000, a new standard, has higher compression rates
and it streams data in waves so that users can choose the resolution they need thereby saving
them unanticipated loss of quality that they can experience with lossy compression.
3.2 Unified Messaging, Voice Information and Vehicle Telematics
We have looked at Unified Messaging, Voice Information services and Vehicle Telematics as a
single category of inter-related services. Whereas these three businesses, in their infancy,
existed as separate businesses, they are now converging into one. Vehicle Telematics is
available from International Imaging Industry Association. Kodak has its own scheme for metadata available at http://www.kodak.com/US/en/developers/tools/02_pmt.jhtml
indistinguishable from mobile services when the same device can be used in and out of a car,
together with Bluetooth devices, as will be the case with services offered by emerging
competitors like Mobile Aria (www.mobilearia.com) in the future. Messaging, information
services and even entertainment can be provided as a single package of service for mobile
customers whether in the consumer segment or the enterprise space.
Messages, whether they are e-mail, SMS, voice mail or fax, are distinguished by their format or
their metadata. They can be converted from one message type to another by changing their
metadata. The conversion of their content from text, like e-mail, into voice mail or vice versa is
enabled by speech-to-text or text-to-speech technologies. A text message does not have to be
read but it can be heard by the recipient.
Speech technologies play a key role in the convergence of these businesses. Vehicle Telematics,
in its early days, was a call center business and offered, besides safety and security services,
location information relevant to drivers. In the future, ATX Technologies
(http://www.atxtechnologies.com/) will continue to offer safety and security services, as a call
center function, while location information will be speech enabled.
The product mix of telematics services providers will expand to include information and message
services in the package they offer to customers.
Multi-media will further blur the distinctions between these businesses. The technology required
to deliver video messages or attachments of music files with messages will not be a whole lot
different from that required to offer entertainment to car owners. The convergence of
entertainment with messaging and information is possible as large media files can be
downloaded (to multi-media player which also reads messages and plays voice information
messages) using Bluetooth or 802.11 technologies.
Some companies are beginning to take advantage of potential economies of scope inherent in
the emerging technologies. Ibasis (www.ibasis.com), provides both unified messaging and
speech enabled information services Comverse includes entertainment as well messaging and
voice enabled information services (www.comverse.com/solutions/index.htm). Telematics Service
providers such as ATX Technologies (http://www.atxtechnologies.com/) and Mobile Aria
(http://www.mobilearia.com/) are working on plans to offer multi-media telematics services over
the next 18 to 24 months. Delphi Automotive Systems
(http://www.delphiauto.com/products/manufacturers/multimedia/) and Visteon
(http://www.visteon.com/technology/automotive/Multi_ICES.html) have developed
equipment that can be used for all these three services including music and video, which will
considerably increase the demand for storage intensive products.
From the supply side, storage technologies help in reaping the economies of scope to the extent
that the data for all these services can be aggregated at a single point in a data store. The costs
of storing messages, as we will see, account for a substantial proportion of the costs of unified
messaging. Technically, its possible to also store voice files and media rich files in a single store
but this is hard to achieve when content is received from multiple sources.
3.2.1 Storage and Unified Messaging
For the sake of expositional clarity, we will discuss the role storage plays in the provision of each
of these services separately beginning with unified messaging. Storage occupies center stage in
web based messaging systems such as those built around the Internet Message Access Protocol
(IMAP), or similar web based e-mail access system. Unlike the more commonly used Post Office
Protocol 3 (POP3), IMAP is a designed to access files, using a web browser, from any location
and at any fixed or wireless device.
Unified messaging requires a single repository to efficiently convert a message from one type,
such as voice mail, to another like e-mail. If the server architecture were retained, messages
would have to be reproduced in both the voice-mail server as well as the e-mail server before e-
mail or a phone client can access them. Wasteful reproduction of messages increases as the
number of channels of access increase.
When messages are centralized in a single repository, they can be readily converted to another
medium. Messages are converted from voice to text and vice versa by speech recognition and
text-to-speech technologies and Optical Character Recognition for conversion of fax messages.
From the consumer end, access from a common repository becomes essential when messages
are accessed from more than one location. Post Office Protocol 3 (POP 3), the most commonly
used protocol, gained currency when the desktop was the only client. Once downloaded, files can
only be accessed locally from the desktop client. Messages can, theoretically, be accessed on-
line from anywhere, with POP 3, if they are saved on the server. Users would, however, find this
inconvenient, unless they have a file management system, which reports on previous activity.
Protocols such as IMAP provide a common visual interface to all files and folders stored on a
message box on the Internet. This is particularly useful when files of different kinds, text, voice,
fax, short messaging and video, are integrated to provide a common view. It has the ability to
provide status report of any previous action undertaken on the file.
Above all, IMAP affords an opportunity to search and retrieve files of his or her choice or a subset
of a file such that the more important sections can be retrieved on bandwidth poor wireless
devices or attachments can be viewed at a later time. When POP3 is used as an e-mail client, all
the pending messages from a server have to be downloaded. Users have no choice but to view
the entire message and the attachments including voice or video files.
By aggregating all messages in a single data store, service providers can economize on a variety
of administrative overheads that are otherwise required to manage messages in their separate
mediums such as directories for voice mailboxes and e-mail servers and the costs of maintaining
user data, operating system and facilities management. Instead, a single directory, the preferred
option is Lightweight Directory Access Protocol (LDAP), is used for the administration of all types
of messages.
In a component based messaging software, the management of the capacity of the data store
takes place independent of other pieces within the messaging system such as the processing
power of servers. The message store can be a RAID, NAS or SAN so that the e-mail storage can
be transferred into another disk when any one of them breaks down. Replication of data ensures
that any break down does not lead to loss of information. Similarly, the disk capacity can be
raised as volumes increase without any interruption in services. Service providers can also offer
classes of services based on the users’ tolerance for downtime.
3.2.2 Storage and Voice Information services
We will now look at the role storage plays in the provision of voice information services.
Traditionally, Interactive Voice Response (IVR) systems were used to automate some of the call
center functions. The tedium of using prompts on a touch-tone phone discouraged the use of IVR
except for a few functions like preliminary instructions.
Customers can now use natural language keywords to search a storehouse of information and
the response is read to them from any telephone. They can be productive in their spare moments,
such as when they are driving or walking, by looking up their e-mail or calendar. Speech
recognition allows them to do this even when they are driving without risking an accident that is a
common experience with the use of a cellular phone.
The ease of retrieval with speech recognition technologies has encouraged companies to offer a
broader range of voice information services. Companies can automate responses to routine
functions such as arrival time for trains, flight departures at airlines or descriptions of promotions.
Just as speech recognition technologies are more convenient than IVR for the consumers, the
introduction of VoiceXML applications lowers the cost of deployment of such services. Whereas
IVR technologies require a separate infrastructure, VoiceXML can be integrated with their text
web infrastructure and their content can be converted into voice.
Voice files are large, ten times the size of an equivalent text file, and their volumes are increasing
with increasing adoption. Storage is required not only to manage large quantities of data but also
its variable demand. An individual company is less likely to fully utilize its infrastructure and would
incur higher costs than if it were to outsource its services. As an example, retail stores receive a
disproportionate number of calls during the Christmas season compared to the rest of the year
when call traffic is more moderate. If the facilities are designed to cater to peak level of demand,
they will be underutilized for the rest of the year. Alternatively companies can invest in a smaller
size infrastructure at the risk of losing goodwill during the holiday season.
Concurrently, the emergence of web services and web servers has separated the function of
application use and the operation of the back-end infrastructure. It is now possible to house the
infrastructure in a data center. When the data storage infrastructure is outsourced, it can begin to
take advantage of the redundancy of the Internet. The access to storage facilities on the Internet
affords an opportunity to scale the size of the infrastructure as the demand grows.
3.2.3 Storage and Telematics
Telematics is composed of several services including safety and security services that have been
the staple of telematics services so far. Increasingly, voice information services are gaining
ground especially because drivers need to access information without holding a wireless device
in their hand. Finally, entertainment services are valued for drivers to use their time.
Storage is required as the product mix of telematics services is increasingly multi-media.
3.3 Location based services
Geographical information can be represented as either vector data or raster data. Vector data is
shown as a set of co-ordinates, X,Y and or Z, that are useful for depicting quantitative information
on maps. Raster data is depicted as cells (bit-mapped) and is useful for graphical representation
of geographical information. Digital representation of satellite imagery, aerial photography is done
in raster mode. Vector data and raster data cannot be used in combination except when they are
juxtaposed.
Currently, the GIS/location based services industry generally utilizes vector data since it
economizes on storage and is useful for measurements. However, raster data is visually
appealing and its use can be effective in industries like the real estate or the travel industry when
a picture of the surroundings can aid customers in their decision-making process. Raster data,
however, requires a great deal of storage space which increases as the resolution improves.
Storage technology will play an increasingly important role in location-based services (referred to
as GIS in the non-commercial world) as satellite imagery of finer resolution becomes available.
Currently, satellite imagery of a resolution as high as one meter is already available12 and
licenses for satellites capable of half a meter resolution have been granted13. Location based
services are potentially possible with pictures of one meter resolution since streets, parking lots,
12 http://www.spaceimaging.com/newsroom/press_kits/factsheet.htm 13 http://www.space.com/businesstechnology/business/satellite_licenses_001218.html
movement of cars becomes visible14. With data of 100 MB per picture15, terabytes of storage are
required for preserving imagery of such high resolution is made available16.
The demand for raster data, satellite imagery and other remote sensing data will grow when such
data is moved faster from their source to the point of consumption. Companies like AXS
Technologies (http://www.axs-tech.com/index_blue.php) offer parallel processing technologies
to retrieve information rapidly from disks. It has also partnered with Inciscent
(www.inciscent.com) to provide wireless access to this data.
Location based services (LBS), as they are known in the wireless world or Geographical
Information Systems (GIS) grow as increasing number of data types are linked to spatial
information. Geographical Information Systems link a variety of data types by tying them to their
spatial characteristics (http://www.cubewerx.com/). For example, the longitude/latitude, zip code
can be utilized to link information on assets of utilities. Due to high costs of storage and
bandwidth, the commercial sector makes limited use of raster data as well as satellite imagery
and aerial photography.
Furthermore, geographical content is created by numerous entities such as city governments,
state governments and federal government. Other information such as yellow pages is received
from telecommunications companies and public and private sectors produce satellite imagery and
aerial photography. This data is now conveniently available from two major clearinghouses-the
Federal Geospatial Data Committee- as well as the Geography Network, established by ESRI
(www.esri.com). Storage area networks help to move data in real time to its consumption point.
Modern day Geographical Information Systems (GIS) visually illustrate location information by
portraying them on maps. Bald numbers from databases and statistical series are rendered
visually on maps for intelligibility. For example a database query on income distribution can be
displayed on maps showing neighborhoods with income classes or statistical data such as
population density can be displayed on maps. Retail chains use such information to determine
the location of each of their stores and correlate them with their sales data. 14 For an illustration of the images possible with one meter resolution, see http://www.estec.esa.nl/ceos99/papers/p154.pdf 15 op cit 16 http://www.pitt.edu/~oduibhne/RIVERS%20OF%20DATA_00.07.06.htm
Satellite imagery plays a variety of roles depending on the nature of the application. In the
commercial world, the most common applications of satellite imagery are for risk assessment in
the insurance industry and asset tracking in the transportation industry. Transportation companies
typically use satellite imagery to aid navigation of their trucks in less familiar territories using
wireless devices (http://www.objectfx.com/). Similarly, telecommunications companies use
location specific data on their cells and central offices for use by their field forces. The insurance
industry uses map data to ascertain susceptibility of a clients’ property to natural disasters for
determining risk and premiums17.
Much paper work in asset management is saved when GIS is utilized. Utilities, for example, have
to respond to enquiries from contractors who need to dig in a particular geographical region. In
the past, utilities had to wade into their paper documents and painstakingly ensure that none of
their assets would be affected. Today, contractors can look at web hosted GIS databases to find
out for themselves whether any harm will be done to utility assets by their digging18.
Wireless location based services have become possible with the advent of Geographical
Positioning Systems and E-911 identification that FCC mandates. In the future, location aware
devices will enable automatic transmission of information triggered by events.
By bringing together scattered data, GIS information systems pave the way to centralizing
information and to take advantage of the scale economies afforded by storage technologies for
archiving and retrieval of the information at relatively lower costs19.
3.4 CUSTOMER RELATIONSHIP MANAGEMENT
Customer relationship management databases have grown from the gigabyte range to the
terabyte range in recent years. Information is now collected from inside and outside the enterprise
on a broader range of parameters and more frequently. Storage technologies are required for
supporting CRM databases in such a scenario; companies such as SAS (www.sas.com),
working in collaboration with EMC and Teradata (www.teradata.com), working in collaboration
17 For a case study, http://www.mapinfo.com/community/free/library/insurance_wp.pdf 18 http://www.mapinfo.com/community/free/library/pocs_casestudy.pdf 19 A sample of location based services can be found at http://www.jlocationservices.com/MarketShowcase/DeployedLocationServices.html#
with LSI Logic (http://www.lsilogic.com/index2.html) are currently the major players in the
larger size CRM databases.
For the larger databases, information is collected not only from an individual department but also
from related divisions in the enterprise. The early CRM databases typically collected data on
orders, billing and collections. They evolved by incorporating related enterprise information from
inter-dependent departments, such as production, thereby enlarging the size of the databases.
CRM databases with operating information such as production, sales and inventory become more
valuable when they are placed in a context. Data on expenditures on music, for example, is
meaningful when seen together with the socio-economic psychographics of individual segments
of the population. Beyond the early adopter stage, companies compile secondary data on
demographics to decipher patterns that are useful in strategic planning.
Also, companies have realized the need to retain their customers instead of incurring the costs of
customer acquisition. They see an advantage in collating historical information on their customers
in order to find patterns in their purchasing behavior that can provide clues to gain their loyalty.
Typically, CRM data is used to identify customers who account for the highest share of
consumption, their responses to campaigns and any product features added to products.
The diversity of information available in CRM databases has given rise to business intelligence,
which is used for deciphering patterns such as segmentation within the customer base. The data
is used to offer customers related products that fit their profile. Their responses to promotions
reveal their price sensitivity. Initially, business intelligence concerned business analysts who used
the information to communicate to senior executives of a company.
Increasingly, companies have realized much greater gains are possible by using business
intelligence to alert their staff in real time. This is particularly true for applications such as logistics
management, i.e., optimizing when conflicting demands are made on time of delivery and cost for
shipping goods. Similarly, seasonal fluctuations in demand for industries such as the travel
industry need quick feedback on demand patterns in specific groups. The dissemination of
information, from centralized repositories, is possible with wireless devices. Mobile staff is alerted
to critical performance data of the company as well as actions of competitors that may be
revealed by say sales data. Business Objects (http://www.businessobjects.com) is one such
company that offers business intelligence services using wireless devices.
The lead times for collection, assimilation and communication of data are lower when data is
collected for real time decision making. Consequently, databases increase in size as data is
collected more frequently.
Data warehouses or corporate information factories, as their larger versions are known, are the
radars of enterprises guiding the movements of their road warriors. The size and complexity data
warehouses behoove their reliable management. Storage area networks provide the redundancy
to manage rising volumes of data and lower the downtime by replicating them at several sites.
3.5 STORAGE AND MEDICAL APPLICATIONS
Fragmentation of workflow is commonplace in the health care industry and is the root cause of
high rates of medical errors, much of the cost escalation and deteriorating quality that has fed
political passions in the USA. The discontinuity exists at four different levels; within departments
of a hospital or a medical group, between branches of a group, across institutions within the
larger healthcare enterprise and geographical regions. The current effort to integrate workflows,
with the help of storage technologies, is within departments of hospitals and medical groups.
The fragmentation of workflows is the cause of paper shuffle, loss of information, time delays as
professionals struggle to collate information and decisions are based on inadequate data as a
result of the discontinuities in the workflow. According to figures collected by the American
Hospital Association, paper work in the medical world takes at least fifty percent of the time on
patient care if not more. Emergency care is the most wasteful with paperwork matching the time
on patient care while skilled nursing takes half the time20
Storage has a potentially vital role in funneling information to a single point, integrating bits of
information, disseminating it to professionals and processes it for decision support. The
companies that lead in leveraging storage technologies for integration of work flows in the health 20 “Patients or Paperwork”, American Hospital Association, 2001
care industry are General Electric Medical Systems
(http://www.gemedicalsystems.com/it_solutions/index.html), in collaboration with EMC, and
Siemens (http://www.smed.com/), Philips
(http://www.medical.philips.com/product_lines/mimit/index.asp) and Agfa
(http://www.agfa.com/healthcare/modelpathpage.php?pageid=1810&type=product), in
collaboration with Storagetek.
One of the first departments to be digitized in the health enterprise has been the radiology
department. Digital imaging yields cost benefits, simply by abandoning the costly processing of
silver halide films, without wrenching restructuring in a health enterprise. The pay-off is larger as
digital archiving is potentially cheaper and the movement of records between departments is
faster. Imaging also crosses departmental boundaries more than any other division of a health
enterprise; it can be requisitioned by outpatient clinics, the operating rooms, intensive care unit,
etc.
In the past, the images had to be moved physically and by trucks when they were transported to
other units within the hospital complex. Storage enabled Picture Archiving and Communications
Systems (PACS) allows health enterprises to deposit the images at a single point and the
associated metadata allows multiple users to access them at several locations. In one
implementation of the project in Cincinnati Children’s Hospital, the time from the end of the
procedure to the sign-off stage was reduced from 37.2 hours in 1999 to 2 hours in 200121.
Beyond radiology, much greater productivity benefits are possible in clinical work flows. At this
stage, automation is harder because work processes are more heterogeneous; information is
filtered from the lens of a particular discipline, individual doctors exercise judgment, nurses and
doctors have differing needs and data required in departments such as intensive care is not the
same as in an operating room. Consequently, data definitions are rife with controversy.
21 Presentation by Neil D Johnson, MD
On the other hand, the payoff from clinical information systems is potentially higher as it has an
impact across the health enterprise. A patient could be treated by a general physician followed by
a specialist and examined in the laboratories. At all these levels, paper patient records have to be
shuffled from one point to another22. Similarly, a patient is cared by doctors and nurses and the
record of treatment has to be submitted to insurance for compensation.
Much of the duplication of paperwork can be eliminated by a clinical repository at the center of it.
An electronic patient record helps to access the current and historical clinical information about a
patient. Sharing of information between health providers is the most compelling administrative
and clinical reason for adopting an electronic health record. Productivity benefits are the next
most important perceived benefit from a centralized patient record (see table).
In addition, data centralized in clinical depositories can become the bedrock for decision-support
systems. Doctors are required to choose between a host of patented and generic drugs; they
have to weigh the trade-offs of effectiveness and the price of the drugs. They also need
information on interactions of drugs with allergies. They need data on historical record of patients’
to judge how a drug will affect them. In addition, they need pharmacological data such as side-
effects of drugs as revealed by the latest research as well the drugs covered by the patients’
health insurance coverage. Computerized Physicians Order Entry (CPOE) used information
about the patient and the condition to narrow down the options that doctors have to make23.
Data traverses a variety of institutions in the health industry when it moves from points of origin to
its destination where it is consumed; information originates or flows to research institutions,
hospitals, home care, medical groups and insurance companies. Information has to be presented
in a variety of ways for each client as it moves from one section of the enterprise to another.
Physicians use pharmacological data, clinical information, and imagery and laboratory results for
diagnosis and prescription. Insurance companies, on the other hand, use similar information for
22 A case study describing the benefits of integration is available at http://www.gemedicalsystems.com/monitor/products/info_sys/qsperin_kaiser.html23 An exhaustive review of the evidence on the impact of CPOE and the opportunity costs of not implementing it can be found at http://www.icsi.org/talist.htm
fraud control. The administrative divisions use the same information for billing, charge capture
and consumer retention purposes. Storage plays a useful role in centralizing information and its
reuse by its presentation in a variety of formats.
Wireless applications will potentially play a vital role reducing medical errors, the costs of
gathering data, quality of care and communicating information retrieved from repositories.
According to one estimate, charge capture alone will add 4% to revenues by saving the paper
work24 and the associated reprocessing that routinely happens because physician’s fees are
missed but not that of the hospital. Other applications include reduction of errors in prescription,
vital sign monitoring and retrieval of data from laboratories25.
The increasing automation of clinical workflows will drive wireless applications moving them from
their initial stage of adoption to widespread usage in the health industry. Medical errors can be
reduced when physicians have a ready reference to the deluge of data on prescriptions, drugs,
patient information, drug interactions, insurance formularies, etc. eProcrates
(www.epocrates.com/products) specializes in applications that provide ready reference to
information on drugs.
Similarly, Allscripts Healthcare solutions (www.allscripts.com/ahcs/index.htm) offers solutions that
allow physicians to use electronic pens to write prescriptions and transmit them directly to
pharmacies, automatically check against the formularies data and for drug interactions. Many
medical errors take place because pharmacies are unable to understand a physician’s
handwriting or they are unable to recall information on interactions of drugs and allergies.
Patients and pharmacies also frequently check back with physicians when drugs are not listed on
an insurance company’s formularies.
24 Quoted from http://www.patientkeeper.com/download/whitepapers/mgma_panel.pdf, page 5. 25 http://www.patientkeeper.com/about/vision.asp
The ability to reference clinical data quickly is another important reason ofr medical errors.
Skyscape (www.skyscape.com/index/) serves as a repository of all manner of clinical data that is
available to doctors on their handheld computers.
3.6 Rich Media Applications
Media is largely stored in an analog format aside from digital media that is created by
companies, such as Einstein TV (www.einstein.tv/uk/index.asp) that began with broadcasting
science documentaries on the Internet. Streaming media technologies can encode the
accumulated analog media assets for reuse on the Internet, as interactive television and video-
on-demand.
Typically, analog media, viewed in movie halls, is cost-effective for mass audiences and tends to
discourage the production of content for smaller audiences such as schools. Einstein TV offers
science documentaries that are otherwise rarely available on analog media.
Analog media is typically viewed at pre-determined points of time whereas digital media can be
seen on-demand. The media is placed on storage devices that are accessible on the Internet so
that the audience can choose the time to view the content.
Video footage created on analog media has a relatively short shelf life. Films are inherently prone
to damage and their preservation requires specialized skills. Consequently, films could not earn
revenue beyond their short life span, which is often inadequate to cover the initial costs of its
production.
By digitizing and accumulating media at a central location in a storage sub-system, storage
technologies have paved the way for reuse of media assets. Reuse of media content is hampered
by the inability to search its sub-components. Video logging technologies, developed by Virage
(www.virage.com) enable the indexing of video content at a more granular level, which facilitates
the retrieval of sub-components of the footage. Computers use metadata or data that recognizes
specific shots to retrieve sections of the footage. Speech recognition makes it possible to use
natural language keyword to search for specific footage. Reuse of old content can take place, for
example, by creating historical content from accumulated footage.
Similarly, reuse of media is constrained by the format in which it is created. Streaming media is
often created for particular media players. Images are created in particular sizes or their color
hues have to be adapted before they are suitable in another situation. Rich media content on
storage devices can be readily adapted, with related conversion technologies, for their reuse26.
At this juncture, storage technologies are used in work group environments and are usually
directly attached to their servers. In a news broadcasting company like CNN, for example,
footage on an event arrives from several different sources and is edited by a team of journalists
working together. Typically, each member of the team uses a workstation to store the footage that
is used as a reference for consultations. Consequently, an enormous amount of superfluous data
is stored on each workstation or moved on the network when it has to be exchanged.
The possibilities of reuse increase with a storage area network which allows content to be
accessed by several different users who could well be spread out geographically. Einstein TV, a
company based in the United Kingdom, for example, is implementing storage area networks that
will help it to distribute its program content to six different European countries in their own
languages. The conversion of the content to the presentation requirements of each of the
standards and the translation of content in different languages is more efficient when content is
drawn from a single source (as profiled in our case study of IBM/Tivoli).
However, the distribution of video content involves complex system integration capabilities that
have not been fully developed. The ability to manage very large files at an affordable price is the
most important consideration for commercial adoption27.
Applications for storage area networks for media management exist outside of the movie industry
in the management of marketing communications. Typically, enterprises have to communicate
with their agencies and need to do it quickly to manage their campaigns. Location shooting can
very well be far away from the place where the footage is processed and it is often used in a third
place. The ability to centralize content at a single point and mirror it at several different locations
is essential for co-ordination between groups working across geographical boundaries28.
26 See case study on Generic Media 27 The issues have been discussed in the case study on IBM/Tivoli 28 This has been documented in our case study on E-Motion (www.emotion.com)
Once the media is produced, enterprises also need to reuse their content. This becomes
necessary when it has to be published not only in the print media but also the web or on films.
Similarly, the content has to be published in collateral which could be in pdf files, excel sheets or
word files. The content has to be made available in a variety of sizes, colors and resolutions.
Large corporations have to be able to do this not only in a single department but several clients
within the enterprise29
Storage enabled wireless applications currently play an insignificant role. However, the
development work for such applications is underway at Virage in partnership with Packet Video
(http://www.pv.com/). In the short-term, wireless applications will be centered on work group
collaboration in studios to prevent interruptions in the editing process. Beyond a two year period,
media content distribution for sports content seems a real possibility.
4 PROSPECTS OF LEADING APPLICATIONS OF STORAGE
Introduction: Rising from the ashes of Internet business models that sought to earn revenue
from advertisements, turnaround for Internet businesses is contingent on consumers’ willingness
to pay. Rich media, voice enabled services, unified messaging, location based services, customer
relationship management are among the applications that the industry, including wireless
telecommunications industry, sees as compelling enough for consumers to pay.
The search for new business models and compelling mobile applications is relentless despite a
string of setbacks in the mobile computing industry. Future prospects in the 3G wireless might
have been roiled by a botched allocation of the radio spectrum, delayed adoption of new
applications, etc., but the leaders in the industry have not lost their verve. Accenture, IBM, Philips,
Sony among large companies and Brience, Airborne Entertainment among start-ups have
wagered audacious new game plans.
Accenture, the reinvented Andersen Consulting, has promoted a bevy of mobile applications that
its Technology Labs has initiated, its venture capital arm has financed and its consulting group
has positioned in the marketplace
(www.accenture.com/xd/xd.asp?it=enWeb&xd=services\technology\tech_efuture.xml) The actual
29 See the case study of Mediabin www.mediabin.com
implementation of new wireless applications is happening in collaboration with leaders in
individual segments such as Microsoft (www.avanade.com/global/ground.zero.asp) and Sony
(www.concadia.com/). The entry of Accenture promises to correct the numerous flaws in
business strategy that jeopardized early launches of mobile applications.
In the consumer space, the entry of Sony and Philips would draw skills in branding and market
entry that the technology industry lacked. Like Accenture, Sony is cultivating ecology of wireless
companies (www.550dmv.com/company.php?cid=44), primarily for rich media entertainment
applications, besides the investments of its companies.
Similarly, Philips can galvanize the rich media space especially because it has decided to
promote standards based (MPEG) applications for its streaming media products
(www.digitalnetworks.philips.com/InformationCenter/PhilipsInternet/FArticle.asp?lArticleId=2059&
lNodeId=997)
IBM, together with its associate companies Tivoli and Ascential Software, has a comprehensive
plan for the wireless space (www-3.ibm.com/pvc/), including embedded devices, which includes
mobile platforms, enterprise applications and related storage infrastructure (www.tivoli.com) and
storage management software (www.ascentialsoftware.com). The hallmark of these inter-related
technologies is the web services model that seems to be the answer to problems of achieving
economies of scale and product differentiation in the application services space.
Brience (www.brience.com), among the new breed of start-ups, has leveraged its adaptive mobile
services platform to offer a range of customized mobile applications to enterprise customers.
Finally, Airborne Entertainment (www.airborne-e.com), another start-up, has a unique model of
micro-entertainment tailored for mobile devices.
Similarly, fresh perspectives are emerging to resuscitate the battered Application Service
Providers companies. The scramble for a piece of the ASP space was misplaced because
players could neither achieve scalability nor product differentiation in the absence of close
relationships with the customers. The ASPs utilized a refurbished client-server architecture that
offered the economies of shared infrastructure to its customers. The rub was that it could neither
customize applications for the needs of its customers nor could it scale in the absence of a
generic application.
An emerging group of players in the mobile applications space offer a generic mobile applications
platform which support components of a variety of applications. The mobile platforms are
generic and can be scaled. Mobile applications are customized to meet the individual needs of
customers. The precise needs of customers are identified by Value Added Resellers who bring
their intimate knowledge of the business processes of the customer as well as system integration
capabilities.
The success of the web services model is contingent on assuaging users apprehensions about
security risks. Over the long run, the success of the web services model will pave the way for
taking advantage of the back-end Internet Infrastructure technologies to reap scale economies.
The web services model allows the sharing of an enterprise database for a variety of applications
used within a company. Load management will be facilitated if its can be distributed over inter-
connected servers and storage devices spread over the Internet.
Coincidentally, the crystallization of several inter-related technologies in storage, especially
centralized storage management, content and digital asset management, transmission of
streaming media, web application services, caching and media players could coalesce at an
inflection point in the near term future. These technologies will lower the inefficiencies in the value
chain that includes data storage, content management or editing and indexing for ease of
retrieval of data, and its subsequent transmission over a network, its conversion into web services
on web application servers and their presentation on a variety of user devices. Computer users,
on their media players, already see some of the impact of these technologies. More is to come….
4.1 Prospects of Digital Photography
Consumer digital photography is potentially a large market since it can substitute for silver halide
films, a well-established mass market, and be one of the leading applications expected to drive
the demand for storage in the near term future. The advantages of digital photography are
• Sharing by means of web storage of photographs, e-mail or disks is instantaneous and does
not require multiple duplications or every time it needs to be viewed.
• Photographs can be previewed on a LCD screen and adjustments can be made before a
shot
Consumers save the relatively high costs of silver-halide films and their processing as well as
indirect costs of disposal of hazardous material.
• Digital photographs can be modified electronically to make them available in a variety of
shapes and sizes.
• Automation of picture taking parameters such as aperture adjustment by in-built software
• Digital cameras can facilitate anytime photography because they can be miniaturized to an
extent where they can fit into Palm devices30.
The reality is that digital photography has been still-born and remained a hobby since 1981 when
it was launched by Cannon and Sony. Personal computers have been responsible for reviving its
prospects. However, personal computers as a medium storage have limitations for the very
simple reason that data stored in them is frequently lost as a result of crashes and virus attacks.
At this stage, digital photography is in a state of infancy; a small percentage of households have
reported ownership of digital cameras or use of photographs. Similarly, e-mail more than Internet
storage of photographs is the preferred means of sharing photographs (see table).
The reasons for low rates of adoption of digital photography are as follows.
• Users of traditional cameras can count on taking shots of a fleeting moment (such as a siblings
quarreling) without losing time on getting a camera ready. Digital cameras have to boot up
before a shot can be taken which can be often too late. Not only is the time taken relatively
long, it is also variable31. Similarly, the time lags between successive photographs is variable32
• The large majority of users still own analog computers and they don’t yet have the option to
preserve their photographs in digital format.
30 (http://www.kodak.com/country/US/en/corp/georgeFisher/dCarpAdp2000.shtml). 31 http://www.mentor.com/embedded/fulfillment/vrtx_dig_cam.pdf, describes the technical reasons for the delays in taking photographs. 32 Op cit
• Ease of use attributes, such as printing, uploading and filing, are not yet available
Historically, ease of use of cameras has been the primary determinant of the diffusion of digital
cameras. Digital photography will undergo its own transformation before it is acceptable to the
mass market. Some of the developments that will make this possible will be
• Creation of kiosks much like the ubiquitous drop-off centers in pharmacy and other retail stores.
Such centers require both printing, uploading and sharing facilities. The leaders in the field
include companies like Pixel Magic Imaging33 and Applied Science Fiction34
• Windows XP has software features, including now Kodak’s Easyshare system that eases the
uploading of photographs to a PC.
• Common printers such as the HP Inkjet can print digital photographs.
• Products such as the Kodak’s mc3, which combines the function of a video recorder, Internet
music player and a still camera, all in a small device will bring to the market the kind of products
that have popular with the young population.
In the past, the photography market has recorded a trend growth rate of six to seven percent.
Between 1997 and 2000, the growth in emerging markets has declined and has offset the
marginal expansion in the developed markets
(www.kodak.com/country/US/en/corp/georgeFisher/pres990427Carp.shtml) so that the
global growth has remained flat. For the purpose of a medium-term forecast, we will assume that
the historical rates of growth will be baseline estimate for the growth in the photography market.
The growth of the personal photography market is expected to rise in the medium-term future
since the rate of diffusion of digital photography will increase aided by the introduction of simpler
devices in the marketplace. Furthermore, the introduction of photo management features in
Windows XP will make digital photography more visible to general mass of consumers.
4.2 Unified Messaging, Voice Information and Telematics
Unified messaging has not been widely accepted despite the perceived convenience of remote
access, for especially mobile professionals to all types of messages, from a single message box
33 http://www.pmimaging.com/default.cfm?href=products_index&article=press_i3_upgrade 34 http://www.asf.com/
and a full listing of messages viewed from a graphical user interface. In reality, users see a
substantial benefit in integrating only fax messages since is received on a separate machine
removed from the desktop35.
Adoption rates for Unified Messaging are presumed to increase as mobile professionals low even
though mobile staff accounts for the majority of staff in SMEs and a significant minority in large
organizations. According to one survey, mobile staff accounts for 10-30% of the staff in large
organizations and 30-80% of the staff in small and medium scale enterprises36. The experience
from actual deployments indicates that price resistance is stiff 37
Software companies have responded to price resistance by leveraging storage technology to
reduce costs. Emerging solutions not only centralize all types of messages in a single store, they
also use the same storehouse for keeping voice information files. The consequent cost reductions
are expected to be substantial (see case study on Tornado Development).
The story of increasing acceptance of voice information services begins with carriers who see a
value in voice portal services. Qwest began the process with its partnership with BeVocal
(www.bevocal.com) followed by AT&T with Tellme (www.tellme.com) and Hey Anita
(www.heyanita.com) with Korea Telecom, Sprint and Net2Phone. Tornado Development began
with Unified Messaging Services with Telekom Malaysia and has recently acquired Global
Crossing. The extension of voice portal business into vehicle telematics and unified messaging is
a short step once carriers agree to deploy voice portal services. More recently, BeVocal acquired
Bell South, the first wire line customer to sign up for voice information services.
The critical performance requirement of voice portals is simply to recognize a wide range of
queries, words and accents. Furthermore, it has to be able to recognize speech even when there
is background noise. Incoming traffic can be very disparate in situations where consumers using
a mobile phone expect answers to any question. By contrast, questions received by an enterprise
would be more uniform; an airline would typically have to respond to queries about flight
schedules. Self-service, with current best practices, is possible with 50% of the consumer calls
35 See the market research report on http://www.unified-msg.com/frames.html 36 See market research study available at http://www.unified-msg.com/frames.html 37 A review of cost issues can be found at www.bcr.com/voicecon/articles/b0008p44d.asp
compared with 14% when speech recognition technologies were introduced. In the enterprise
space, where queries are more standard, self-service is possible with 97% of the calls.
Demand for voice portals is driven by the need to automate call center functions. Shortages of
call center personnel and their increasing costs will impel increasing recourse to speech
recognition technologies. Benchmark Portal, a giant data warehouse on call centers, recently
collated information on the trends in the call center labor market; its data shows that 22% of the
call centers had to struggle to recruit staff, 50% of the staff was performing at less than the
expected levels and 63% of the call centers had already raised salaries or were planning to do
so38.
Automation of call center functions promises steep increase in productivity and cost reduction.
According to Benchmark Portal, the costs of a call processed by a live agent are $ 1.50 per call
while an automated service costs $ 0.25 per call. The sunk costs in Interactive Voice Response
systems and legacy systems and the unproven nature of speech recognition technologies
prevented companies making a transition to automation. However, relocation of call center
functions to offshore locations, especially India, will put increasing pressure to cut costs on call
centers (www.teleworkingindia.com/Senario.htm#Many%20kinds%20of%20I.T.).
Not all call center functions can be automated since customers will have queries that require
information processing. As an example, customers could call a call center at a
telecommunications company and request for information on plans which would typically require
comparative information on promotions, costs and benefits of features, bundling options available
with them and so on. Call center representatives can help in evaluating the trade-offs of each
plan. On the other hand, information such as flight schedules, weather reports are easily
automated.
The business landscape for vehicle telematics has been radically transformed from the time
safety and security was the primary need. Increasingly, voice information services and more
importantly entertainment will constitute an increasing share of revenues earned from vehicle
telematics. These applications are media rich and require storage technologies to support them. 38 Improving Call Center Performance through Optimized Site Selection by Dr. John Anton et al, June 2001.
A transition from analog to digital systems is an important reason for the intensified competition in
the vehicle telematics marketplace. The incumbent, Onstar, could virtually monopolize the market
place as long as safety and security was the only viable service that could be offered. Digital
systems work better for richer applications like entertainment. Early signs of the radical
transformation of the industry are the satellite entertainment services from XM Radio
(http://www.xmradio.com) introduced by General Motors for Cadillac DeVilles and Sevilles, in
November 2001. Similar services are also available from Sirius Satellite Radio
(http://www.siriusradio.com/)
New companies such as Wingcast and Mobile Aria, besides new independent service providers
such as Automobile Association of America, are poised to introduce a wide range of new
services. Emerging players are using dual mode devices since the coverage of digital networks
is still less than half of analog networks; AMPS (advanced mobile phone service) is the analog
system which has 90% coverage in the USA while digital systems have 40% coverage.
While security and safety services were ostensibly the most important services, the actual call
pattern showed that many of the queries were for location based services. Figures released by
ATX Technologies
(http://www.atxtechnologies.com/responseops/newsletter_sq00.asp#handling) show that
only 2% of calls received are emergency calls that necessarily require human intervention. The
large majority of the calls request navigation assistance, roadside assistance and non-emergency
911 calls, usually by drivers who have lost their way and want to get back on track. Much of this
traffic can be automated by speech recognition technologies.
Vehicle telematics, today, has a small base of users that has grown slowly because the pricing is
not attractive. Of a total of 1.7 million users today, 1.3 million subscribe to Onstar services and
the remaining to ATX Technologies. Current market research indicates that consumers are
unwilling to pay any more than a hundred dollars for telematics services39. Onstar currently
charges $199 for its basic plan of safety and security services and $399 for the premium
39 Quoted from “The Current Market for Telematics: Great Products searching for demand” by M Scott Ulnick and William Haupricht of Ducker Worldwide Inc
package, which includes information services. Some of the cost reduction required to lower prices
will be achieved by speech recognition technologies.
A more serious problem with analog technologies, in the perception of emerging players, has
been the product mix that was not in line with market demand. Security services are highly valued
but are infrequently utilized by customers. Information services such as weather, stock, flight and
sport information have both low value and low frequency of usage. Customers look for car
entertainment above all and value hands free voice telephony, instant messaging, voice-enabled
e-mail and the ability to track their buddies and be able to chat with them. Vertical business
applications come, business finders, real time traffic updates and network updated calendar rank
as the second most important set of services required by customers40.
Two major developments are about to change the product mix that will be offered in the future.
Firstly, the launch of Blue tooth enabled technologies that help to synchronize between the car
system and cellular phones. The phone can be docked inside the car and all information is
received hands free by voice messages relayed by the car speaker41.
The greater convenience of voice-enabled services will be a boon to the growth of vertical
business applications that mobile professionals such as real estate brokers and salespeople
require.
Secondly, emerging players in the telematics industry are taking advantage of declining costs of
802.11 A&B wireless LANs to inter-link home computing and telematics. Customers can take
advantage of its high bandwidth of 802.11 (50MB/sec for 802.11 A over a 100 meter distance and
11 MB/sec for B over 200 meters. Data or music and other entertainment files can be
downloaded from the home computer to the car for use in transit.
ATX Technologies, one of the incumbents, on the other hand, is taking a more conservative view
of the market. It does see a market for streaming media and Unified Messaging including e-mail,
voice mail and instant messaging imaging as well as voice-enabled information services. Its focus
remains services specific to vehicles such as vehicle customer relationship management for
40 Rankings quoted from “Perspectives on Telematics”, by Sachal Gidwani, November 7th 2001, Delphi Automotive 41 http://www.daimlerchrysler.de/index_e.htm
OEMs. In any case, even such a minimalist strategy will mean and increase in demand for
storage for media rich content.
In sum, the emerging vehicle telematics market will not only increase in size, it will also raise the
demand for storage services for the supply of new services.
4.3 Business Intelligence
We will estimate the demand that business intelligence databases will create for storage on the
basis of three key assumptions. One, the growth in the demand for business intelligence
databases is driven by overall rates of growth in IT expenditures; we will assume the trend growth
rate in IT expenditures ignoring the abnormal growth in IT expenditure that took place in 1999 and
2000 driven by the euphoria following the launch of the web browser in 1995 as well as the
availability of a surfeit of cheap financing from vendors, junk bond markets, venture capital and
IPOs.
According to a review of IT expenditures over forty years completed by Morgan Stanley Dean
Witter42, the average rate of growth of IT expenditures in the US economy has been 13% percent
per annum. In 1999, it rose to 18% growth over 1998 and in 2000 it rose by another 23% over the
previous year. The excesses are palpable from the fact that 70% of venture capital financing and
56% of the IPOs in twenty-one years took place in these two years. We will assume that the trend
rate of growth will be restored in 2002 and 2003. In 2001, the estimated growth rate for IT
spending is the current consensus rate of 7%.
Secondly, expenditures in CRM/Business intelligence will translate into demand for storage only
when companies plan to analyze data for decision support. The latest industry survey undertaken
by the Data Warehousing Institute (DWI) 43confirms that two-thirds of the companies deploying
analytical databases prefer a centralized approach required for hosting data warehouses on
storage systems.
When CRM databases are used for strategic purposes, their scope expands to the entire
enterprise. Also, the size of the databases grows as information from secondary sources of
42 Technology and the Economy: An attempt at Pattern Recognition, April 2001 43 (http://www.dw-institute.com/industrystudy2000/Chapter3/chapter3.html)
information is also gathered. Companies are clearly making a transition from the early adopter
stage of CRM to more advanced functions of business intelligence. The DWI survey finds that
59% of the companies planned for an enterprise-wide CRM system.
Unsurprisingly, the size of the databases is expected to increase as companies plan to collect
not only operational data but also contextual data to identify significant patterns. According to the
DWI survey, the large majority of the respondents have CRM databases with less than 100,000
records. Over the next 12 months, 53% of the companies are planning to maintain databases of
sizes ranging between 100,000 and 50 million records. Currently, the most popular pieces of
information are account and sales and purchase information and external information such as
demographic and industry information is also collected extensively. Over the next 12 months,
interest will shift to on on-line sales, web registration data, responses to sales campaigns and
multi-media data.
Our third assumption is that legacy CRM databases will, created by early adopters, will migrate to
business intelligence. The early adopters, according to the DWI survey, constitute 25% of the
population and all respondents except 9% will invest in CRM.
A reasonable assumption for our estimate for our forecasts is that about half of all enterprises in
USA will complete the process of adopting CRM technologies required for analytical purposes.
Since two thirds of these prefer a centralized approach, about a third will of all enterprises take to
business intelligence over the next two to three years. We will assume that the early adopters will
have the largest databases of 50 million records while the rest of the 75% will have an average of
1 million records.
4.4 Prospects of Location Based Services
Location based services have been seen as synonymous with consumer applications such as
information on restaurants, driving directions, traffic data and so on. Based on our interviews,
most carriers are unwilling, in the near future, to step beyond consumer applications although the
software industry has the capability to deploy richer enterprise applications. However, Nextel has
had a great deal of success in expanding the market for business applications including location
based services.
Lately, a bevy of established players in the Geographical Information Systems (GIS) into wireless
LBS has brought financial strength and a wealth of geographical information that can potentially
be used to offer richer enterprise applications. Some of the prominent names are Autodesk
(www.autodesk.com), which brings the advantage of its large pool of spatial data useful to the
engineering industry. ESRI (http://www.esri.com/news/arcuser/0401/bunchls.html) funnels a wide
variety of geographical information from its geography network that can be useful for enterprise
applications. Intergraph has set up a subsidiary Intelliwhere (www.intelliwhere.com) for wireless
location based applications. MapInfo (www.mapinfo.com) has long provided asset management
solutions to the telecommunications industry and customer relationship management.
All these companies accord a higher priority to enterprise applications; their current pre-
occupation with consumer applications reflects the preferences of carriers. Autodesk has
developed location based services for the engineering industry and its horizontal applications are
consumer applications offered to carriers. MapInfo offers vertical applications for CRM and asset
management applications while its horizontal applications are offered to the consumer market.
Space Machine is focused on applications for field forces but it has applications that can provide
location based information on restaurant menus.
The adoption of these consumer applications has been stymied by the absence of technology of
identifying a customer’s location. However, increasing availability of miniature GPS devices and
implementation of the second phase of E 911 (see table on the pace of growth of cell-id
technology) will ease this constraint Even in an optimistic scenario of rapid adoption following the
acceptance of E-911 or Geographical Positioning Systems (GPS), the size of the consumer
location based services market is smaller than the short text messaging services market44. From
44 The data on the market for location based services is reviewed in The Economist, October 13th-19th 2001
a storage perspective, the volumes of information will be small since most consumers will require
map-based information without imagery.
Enterprise location based services, on the other hand, are potentially large because they add
value to existing corporate databases such as CRM, asset management databases used by field
forces and so on. Location based services are logical extensions to wireless functions lately
added to widely adopted CRM software such as that available from Siebel.
Furthermore, enterprise databases have a need for large volumes of data such as imagery
required in the real estate industry (e.g. www.realtor.com) or the insurance industry. While vector
data is adequate for most consumer applications, the enterprise market requires raster data for
mobile professionals to visualize data for effective maintenance such as repairs of utility assets.
Satellite imaging, for example, is used in the utilities and the telecommunications industry
(http://www.spaceimaging.com/newsroom/releases/2000/apps-util.htm)
In addition, enterprise users will need to access data more frequently during the course of a day
instead of sporadic use of databases by consumers. Consequently, location identification
technologies like Geographical Positioning Systems (GPS) will play a catalytic role since they do
not require manual entry of zip code information, as is the current practice for consumer
applications.
4.5 Medical Applications
Public policy has an overriding influence on the business environment of the health industry and
this is also true for deployment of information technology. The Health Insurance Portability &
Accountability Act of 1996 (August 21) (HIPAA)45 will have an overarching influence on the
management of health records including their format, storage and utilization. The long debate on
issues of privacy, inextricable from standards of identification of individuals and institutions, in the
45 Background can be found at (http://www.hipaadvisory.com/regs/HIPAAprimer1.htm).
health industry has been settled and has paved the way for accelerated deployment of
information technology including storage technologies.
The section 1173 (b) of HIPAA mandates unique identifiers for individuals, employers, health plan
and health care provider46. Aggregation of information hinges on the ability to identify the patient,
the provider and the payer. The medical history of a patient can be retrieved even if he or she
changes location, health plans or medical groups. All related information including medical
images, patient data, vital statistics and historical information relating to each patient can be
collated. The communication of data is easier since all this information will be available from a
single point by using the metadata for the participants.
The assimilation of information for the purposes of interpretation of the data will be possible
because codes will be used for each procedure, disease type, the symptoms, related health
problems and drugs prescribed. This coding can become the basis of the metadata that will be
useful to find disease patterns and to pinpoint errors in diagnosis, fraud control, etc47.
For a long time, the vocabulary for data definitions in the health industry has been contentious
and varied a great deal between institutions. For example, individual institutions could well have
different scales to measure pain. The section 1173 (c) mandates standard data definitions which
can be used across institutions. The aggregation of data from numerous institutions creates the
conditions for aggregation of data on a very large scale and its use for decision support.
The HIPAA also creates the conditions for automation of information management processes
beyond digital imaging that has been the case so far. Increasingly, health enterprises will see an
advantage in extending automation to clinical information system and will include patient records,
laboratory results and clinical workflows. The utilization of wireless technologies will see a
concurrent increase for collection of vital signs information, dictation of physicians’ notes and
point-of-care data use increases. 46 http://www.hipaadvisory.com/regs/law/1173.htm 47 See http://www.hipaadvisory.com/regs/Regs_in_PDF/finaltrans.pdf for information on data types and identifiers.
Actual implementation of the HIPAA regulations is planned to be completed by 200348. Past
experience suggests that postponement of the schedules is common and is anticipated again49 .
The fact remains that the progress in the larger hospitals, with more than 400 beds, is
unmistakable and extends to identifiers50. Many of the other health enterprises are taking a wait
and see approach and will change when the pace of change in the industry picks up.
The heightened interest in wireless devices and ASPs is another indication that centralized
management of data is well underway. The rate of adoption of wireless devices and ASPs is
expected to be doubled over the next two years51. Similarly, computerized patient records have
already been completed in 13% of health enterprises and are under implementation or planned
for implementation half of the enterprises surveyed52
5 COMPETITIVE ISSUES
5.1 Digital Photography
Ease of use has been the lynchpin of competitive strategies in the photography business. Digital
photography industry, still in its early stage of development, does not have the rudimentary
infrastructure for development of photographs after they have been shot. Digital photography
kiosks will provide the convenience of processing, printing, uploading and distributing
photographs. Companies with the better software and storage infrastructure for the kiosks will be
the winners.
Traditional photography continues to retain the loyalty of the large majority of customers. Picture
quality and ease of use of traditional cameras are strengths that digital cameras will find hard to
match in the near term or even over the longer run. However, digital photography is decidedly the
technology of choice for all the subsequent processes for production of photographs and their
48 http://www.hipaadvisory.com/news/compliancecal.htm49 http://www.himss.org/content/files/hipaa_survey_fall2001_9027.pdf 50 op cit 51 http://www.himss.org/2001Survey/Main.htm 52 http://www.himss.org/2001Survey/Main.htm
preservation and distribution. The companies that marry the advantages of digital photography
with traditional photography will be the winners.
Currently, most needs for storage of photographs is met by removable storage media such as
flash cards, zip disks or CD-ROMs. Internet storage is not yet the preferred medium of storage at
this point of time because of unsatisfactory compression technologies. We can expect companies
to utilize JPEG 2000 and similar technologies to overcome this problem. They will take advantage
of the ability of storage to provide photographic output in a variety of shapes and forms to
increase the size of its client base.
Finally, Meta data is integral to the success of any player in the digital photography industry. The
companies that are able to propose universally acceptable data definitions of photographic
information will be the winners in the future.
5.1.1 Applied Science Fiction’s (ASF™)
Applied Science Fiction’s (ASF™) (www.asf.com) is our choice of a company that has launched a
disruptive technology that will reshape the digital photography industry and accelerate the
currently slow rates of adoption in the digital photography industry. The hallmark of its strategy is
that it brings it brings the ease of use of traditional photography to digital photography and the
ease of storage and editing of digital photography to traditional photography.
ASF™’s innovations take the beaten path of adding ease of use to photo processing of traditional
photographs. Yet it brings to silver halide films unforeseen benefits of digital photography—
storage on a disk, real-time editing, and ability to view images on any device as well as
transmission of images over networks all in minutes.
Users of the 35 mm camera can have the best of the analog and the digital world because ASF’s
Digital PIC™ develops a standard film and creates an Extended Range Digital Negative™
(XRDN™) on a CD instead of a silver-halide film (film negatives). Hazardous chemicals are not
used for processing, as is the case with analog film processing. The XRDN has besides all the
attributes of a digitally shot photograph, such as ease of duplication, printing, editing and storage,
a core file for each image that provides an open pathway to future file formats. Unlike existing
photo CDs, the XRDN is intended to provide the benefits of the film negative plus the advantages
of the digital environment. ASF™ will manufacture the Digital PIC capture engine and sell it to
OEM integrators.
ASF’ corporate strategy is based on the premise that the inherent technical limitations of digital
cameras have inhibited their diffusion since 1981 when they were first launched by Sony and
Canon. Typically, a digital camera converts light into electrons by either a CCD (Charge Coupled
Device), for higher quality and more expensive images, or CMOS (Complimentary Metal-Oxide
Semi-conductor) for lower quality and inexpensive images. An analog to digital converter then
transforms the electrical signals into digital information. These conversions add to the “boot to
shoot” time or the interval between consecutive shots. Most professional or experienced amateur
photographers are unwilling to tolerate these delays since they lose their most important
opportunities.
The other limitation of digital cameras is their need for voluminous storage, which grows as the
resolution improves. For a resolution of an equivalent of 18 Megapixels (2000 x 3000 x 3
channels) that an XRDN provides for a film roll of 24 shots, storage needs of a digital camera are
18 MB, which increases to 30 MB when the pictures are scanned (12 bit depth). ASF’s Digital
PIC technology allows users to keep their files in three different sizes—300K for JPEG
compressed files, 2 ½ MB as lossy compressed files and 121/2 MB as lossless uncompressed
files which can be adapted to future technologies as they emerge.
ASF’s innovations set out to cater to the latent demand for digital imaging from users of traditional
cameras. The market for analog cameras is already very large and accounts for all but 3 to 8
percent (estimates vary) of all cameras or 1.04 billion units excluding 300 million single use
cameras and generate 80 billion images a year. Users of a 35 mm camera value its ease of
clicking and want the advantages of sharing, file management and preservation of digital
photography.
The thrust of its operational strategy is the adaptability of the Digital PIC process. Digital PIC can
be configured to fit within traditional and non-traditional photofinishing markets. A Digital PIC
kiosk rendition will be launched at Christmas time in 2001 and will grow in numbers in 2002. The
concept of a kiosk is much like the ubiquitous drop-off centers available at pharmacy stores for
traditional photography.
The kiosks will bring three main benefits, storage on a removable CD, file management and
printing, that will enhance the usability of digital photography. Currently, most users download
their digital pictures from a digital camera, by a tedious process, to a personal computer for
storage. When stored on PCs, users run a high risk of losing their photographs. Alternatively,
customers store their photographs on flash cards, which have a limited capacity.
ASF’s technologies will provide the photographs on CDs, which can be viewed on not only a PC
but also on a DVD player or a game machine. Users can upload only to their PC, but they can
only show the pictures in VideoCD format on a DVD player and game console. Customers will
have the option to receive a spare copy as a back up. Kiosks will also offer the option of
uploading photographs to an on-line storage device.
File management or the metadata, which identifies the photographs or their context, is another
benefit that will be available with ASF’s technologies. Users can store their pictures in specific
folders each of them describing a context. Current technologies number the photographs and
replace previously saved pictures stored on a device.
The CDs will also not be identified using an adhesive label, as this could lead to future media
degradation that happens when the paper interacts with the glue. Instead, the title of a CD and
the thumbnails of the images on the CD are printed onto the CD without using a damaging print
technology.
Kiosks will provide printing facilities to customers. Unlike pictures printed on home printers, kiosks
will provide prints that last much longer and have a higher quality.
In summary, the Digital PIC kiosk model will create a bridge to digital while providing consumers
with the convenience, control, flexibility and choice that are demanded in today’s fast-paced
digital environment.
5.1.2 Pixel Magic Imaging (www.pmimaging.com)
Pixel Magic Imaging plays a complimentary role to that of Applied Science Fiction in facilitating
ease of use in digital photography industry. It franchises kiosks, which also buy its photo
processing equipment, and provides the storage infrastructure customized for image
management needs. The retail stores, however, are not under any obligation to use Pixel’s brand
name. It also provides local wireless to users while they are shopping in a retail store besides
web access.
5.2 Unified Messaging, Voice Information and Telematics
Two factors, bundling and personalization as well as pricing, will be decisive in the competitive
dynamics of this segment of the wireless industry and storage technologies will help companies in
achieving both these goals. The bundling of services will help in product differentiation and price
reduction as well as in raising their joint utility. Unified messaging alone may not be very useful to
mobile professionals but its value is more when it is voice enabled and is available in combination
with information services.
Service providers will have a choice of bundles of services that they offer to their customers. This
could range from entertainment only or safety and security services only to a combination of
information, messaging, security services and entertainment. The precise mix that customers
request will depend on their needs and their comfort level with devices. The winners will be the
service providers who are able to provide the broad mix of services and to tailor them for
individual customers. Storage technologies will help to centralize the content at one point and to
manage their delivery to a variety of customers.
The other critical factor in the competitive dynamics would be to reap the economies of scope
from aggregating services. By offering a bundle of services, companies will be able to not only
reduce the cost of marketing but also to leverage storage technologies to reduce the cost of
managing data.
5.2.1 Tornado Development (www.tornadodevelopment.com/)
Tornado Development is our choice of a company that has made innovative use of storage
technologies to push the envelope for wireless applications. Its architecture is designed to not
only offer a variety of bundles of services but also to reduce costs by centralizing information in a
repository.
It develops enhanced services software including e-mail, voice mail, faxing, SMS, unified
messaging and Multimedia Messaging Services (MMS). These services are licensed from
Tornado and deployed on carriers’ networks. The same applications can also be used for voice
portals, which is the business Tornado is currently entering. The cornerstone of Tornado’s
strategy is to develop software that substantially reduces the costs of messaging, unified
messaging and voice portals by lowering the investment on storage.
The carrier hosts Tornado’s product, Tornado Messenger, with the message store on storage
area networks to take advantage of their scalability. StorageApps
(http://www.storageapps.com/), recently acquired by Hewlett Packard, is one of its important
storage partners besides EMC. The messages are stored on SANs while the associated
metadata is kept on network-attached storage.
The centerpiece of Tornado’s strategy, to reduce costs, involves moving beyond IMAP/LDAP type
of solutions. While IMAP/LDAP solutions have a common repository for storing messages,
Tornado’s view on these types of messaging solutions is that the messages retain their
metacontent (the context of the message such as attachments, date, etc.) and their envelope
(routing information, subject, sender and receiver). Thus, there are compartments for e-mail;
voice messages, SMS and messages from one compartment has to be reproduced in the others
before they can be accessed by all means such as phone, web browser or an e-mail client. The
multiplicity of compartments drives up the costs of storage.
Conversely, Tornado’s architecture allows the merger of all the separate compartments within the
message box. Incoming messages are stripped of their metacontent and the envelope or the
format (neutralized i.e. overheads are removed), which are stored separately in an Oracle or any
other database. Each message is routed to a storage sub-system. Without the complexity of the
metacontent and the format, message content can be compressed to a tenth of its initial volume.
Storage space can, therefore, be considerably reduced. Tornado reports a cost reduction of 25%
with its software.
The Lightweight Layer within the Tornado Messenger Message Server plays a pivotal role in
reducing costs of storage. It neutralizes messages (separates from their metacontent), such that
they are all reduced to data and can be stored in a single data repository. Developed as a module
with C++ and Java, the Lightweight layer translates messages from one format to another, in real
time, so that they can be sent and received by any means of access.
The Database Layer of the Tornado Message Server, where the administrative information
resides, identifies and catalogs the stored messages. It also plays the role of routing the
messages to the device where it is finally received. The Database Layer works with the Remote
Access Communication Layer, at the edge of the network, which acts as a gateway between
Tornado’s IP network and the PSTN.
The Lightweight Layer, which incorporates a data link layer for database connectivity, can be
connected to existing e-mail or voice-mail server. Customers of unified messaging services
typically don’t want conversions from one message type to all others. They look to progressively
increase the message types that they want to unify over time. Such clients have the option to
migrate without scraping their existing infrastructure.
Tornado’s marketing strategy reflects its architecture. In a few cases, such as Telekom Malaysia,
it has been able to sell unified communications software that includes voice, fax and text which
can be accessed from any device whether phone, e-mail or web browser.
The requirements of Tornado’s customers range from e-mail only and voice mail only systems, to
full UM and mobile UM offerings (e-mail and SMS). The architecture allows them to choose single
message systems and integrates other message types according to customer needs.
Most of Tornado’s customers reside in East Asian countries and in Europe. In the USA, it has
launched its services with Global Crossing and has entered, or is slated to enter, into system
trials with 3 U.S. wireless carriers.
A common message box will be the foundation on which Tornado will, over the next 6 to 18
months, plans to add voice portal services, and Multimedia Messaging Services (MMS) (video,
picture messaging, audio clips) and intelligence to messaging. Users will have the ability to sort
messages on their criteria of priority or reject spam as they define it. The system keeps track of
patterns of messaging and users predisposition to open any type. Based on the profile of the
user, the intelligence function will add automation in acceptance of messages and prompts for
retrieval.
5.2.2 BeVOCAL (http://www.bevocal.com/index.html)
BeVocal is our choice of a company that has found ways to increase the rate of adoption of voice
portal services. Although it uses content from its partners and does not directly use storage
technologies, it has made advances in expanding the market for voice information services.
BeVocal, a voice portal technology company, came into the limelight when it achieved the crucial
breakthrough of acquiring Qwest as a customer. Other voice portal companies, Tellme with AT&T
and more recently HeyAnita with Sprint, followed. BeVocal, however, maintained the lead by
confirming, in October 2001, that it had acquired Bell South, the first wire line carrier and regional
Bell Operating Company to adopt voice portal services. The partnership with Bell South signals
the end of the early adoption stage for voice portal companies and paves the way for reaping the
benefits of economies of scale.
BeVocal does not expect revenue growth from wire line carriers to be as rapid as with the
wireless carriers. On the other hand, Bell South brings the advantage of a larger market, which
will grow with vertical product differentiation.
While wireless carriers cater mainly to the consumer market, Bell South’s market also includes
enterprises and call centers. BeVocal plans to work with system integrators and third-party
certified developers (developers on the BeVocal Café (http://cafe.bevocal.com)) to customize
services for the carriers in a variety of vertical markets in the enterprise space. For the consumer
market, new innovations will include e-centers, the equivalent of a public phone for voice
information services. The demand for voice portal services is expected to grow in call centers
since relocation of such services to South Asia has heightened pressures for automation.
BeVocal also plans to extend its portfolio of products to include, besides voice information and
voice dialing services, new services such as voice messaging. Its voice information services
include location and travel information (driving directions, business finder, weather, traffic and
flight information), information services (stock-quotes, news, sports) and entertainment (movies,
TV dramas, horoscopes and lottery). Increasing competition is expected to lower the margins for
these services. Higher margins are expected from introduction of newer products such as voice
dialing and e-mail communicated by voice.
For the management of fluctuating traffic, BeVocal does not plan to take advantage of the
redundancies typically available when the solutions are hosted on data centers. Instead, it has
developed proprietary software, which helps to balance loads from different types of traffic. To the
extent variations in traffic for each piece of information service is known and moves in contrary
directions, the software is able to change the routing of traffic such that each port is optimally
utilized.
BeVocal’s platform can accommodate both VoiceXML and speech objects and either of them
allow componentization of its applications. Storage management is not a concern as a result of
the layered software architecture. At this point of time, it did not report storage as a significant
barrier to performance.
5.3 CRM/Business Intelligence
Data warehouses are used for strategic as well as tactical decision-making. A typical strategic
use of a data warehouse is the analysis of customer churn or the analysis of fraud, which does
not involve extensive participation in an enterprise. Furthermore, predictive models for such
applications are tailored for specific enterprises. The picture changes with tactical decision
support such as pricing of rooms in tourist resorts during the season. Information is fed,
processed and communicated in real time for such applications. Data flows are unpredictable and
the variations have to be managed by an appropriate design of storage systems. Wireless
technologies are critical for receiving information and to communicate decisions about pricing.
Data warehouses help in collation of information, reducing time lags in processing of data and its
communication. It is harder to cross-reference information when it resides in isolated data marts.
Similarly, its hard to communicate information to a group when each person is required to access
different data marts. Companies such as Business Objects (www.businessobjects), Information
Builders (www.informationbuilders.com) and especially Teradata (www.teradata.com) and
(www.sas.com) have transformed the competitive landscape of the Customer Relationship
Management (CRM) marketplace by cutting short the time lags in decision-making.
In a data warehouse environment, information is utilized to speed up decision-making processes.
This is despite the fact that larger volumes of data need to be gathered from a growing number of
touchpoints and centralized in single repository (they can be distributed in a web environment). A
richer taxonomy of data needs to be assimilated and presented graphically as key indicators that
users of the information can act upon in real time. Wireless technologies play a complementary
role since they can be used to receive prompts for decision-making.
Some of the key factors influencing the competitive strengths of a CRM solutions company are
• The need to assimilate information from diverse sources requires the ability to convert data
from diverse systems. A company may, for example, receive information from its suppliers,
retail stores, handheld devices, web site and e-mail. Besides operational data, information
about market developments, such as leading indicators, is received from information
suppliers such as Dun and Bradsheet. In order to aggregate information, CRM companies
have to be able to develop interfaces that allow the flow of information from all such sources
to a central repository.
• Traditional CRM systems utilize Online Transaction Processing (OLTP) systems for data
recording and data processing. OLTP, which uses a relational database, is efficient for
recording information since data elements for each entity (such as product, orders, etc.) is
recorded in a separate table (consequently information such as numerical identification does
not have to be reentered). OLTP is useful for simple queries such as comparison of inventory
and sales. On the other hand, SQL queries become increasingly complex, as queries require
cross-reference to several different entities and time for retrieving data increases. Business
intelligence software is designed for parallel processing of data and parses data concurrently
across several entities. Typically, the tools used for analysis are called Online Analytical
Processing (OLAP) which have the ability to respond to more complex queries53
• Large data warehouses are valuable to an enterprise when the “woods in the trees”
messages from masses of data can be reduced to key indicators and visually illustrated. This
is particularly necessary when information has to be communicated to the broad majority of
53 A description of OLAP can be found at www.olapcouncil.org/research/whtpaply.htm
employees who have very little time to absorb information. In the past, data analysis was
done by drawing data from relational databases that required familiarity with SQL queries.
Emerging data processing technologies enable presentation of data in the spreadsheet
format that most professionals understand54.
• Collection of information at source and communication of information by mobile devices
undergoes change with the processing of large volumes of information. Currently,
synchronization is the most popular means to update information available in a database.
Typically, information is stored or retrieved manually by a handheld device or a laptop onto a
personal computer or a server after an agreed interval of time. Synchronization is preferred
for its low costs. Business Intelligence data is best communicated automatically as alerts or
when data is pushed to the beneficiary of the information. Alerts are triggered by events such
as shortfalls in inventory, which is communicated to salespeople who need to act on the
information.
• Storage technologies are critical for the management of downtime and retrieval time on data
warehouses. By replicating the data on several storage devices, especially when the number
of users rises, time lags in retrieval can be lowered. Similarly, downtime can be lowered with
redundancy that is possible with replication of data.
54 For a demonstration of one such product from ProClarity, see
5.3.1 Teradata (www.teradata.com)
Teradata is a leader in business intelligence data warehouses that shorten the lead times for
collection, processing and dissemination of information for multi-terabyte size databases. Its
business intelligence software goes beyond the use of data warehouses for the analysis of data
for strategic purposes. Instead, its Active Warehouse can be deployed for near real time decision
making by large number of users within an enterprise as well as partners within the extended
enterprise. It has designed its proprietary data warehouse supported by its unique file system to
manage storage required for I/O that is entirely unpredictable.
In terms of sheer size of the data warehouse, a typical application of a Teradata warehouse is the
case of 3M, a multi-national company which has operations in 200 countries, manufactures
500,000 products, employs 75,000 people and has a turnover of $ 15 billion. Before the
implementation of the data warehouse, data collection and reporting was confined to each
department and customers and partners were, for all practical purposes, dealing with 50 different
companies. A global, enterprise wide data warehouse integrated all these disparate databases
into a single warehouse which can be searched, queried, sorted by a web browser. In addition,
the data warehouse integrates information from D&B Worldbase which includes data from 50
million businesses and has worldwide customer information and market channel information.
Altogether, 30,000 users access the data warehouse each day with 15 million hits.
The nature of the decision support that Active Data warehouse can provide is illustrated by the
case of choices that truck owners have to make. Truck routes are differentiated by those that
have their destination as the hubs and others carrying freight from hubs to points where packages
are received. A truck driver has to decide between the options of fully loading the vehicle against
the time loss by waiting for additional packages to arrive. These decisions have to be made in
real time. Information processing required for the decisions would be random as it would involve
estimating time of arrival, distances covered by trucks and so on.
For its storage solutions, Teradata made a deliberate choice not to use storage area networks. It
has, instead, decided to have directly attached storage in order to guarantee the performance for
query time. In a storage area network, the volume of traffic is entirely unpredictable since the
number of nodes is very large so that conflicting demands are made on the capacity of storage
area networks. Teradata, on the other hand, uses hashing techniques to guarantee a fixed
amount of storage for a given level of processing needs. Storage capacity is not physically
allocated for specific needs but is logically reserved for specific processing requirements.
Hashing algorithms reallocate capacity with the ebb and flow of traffic.
Teradata guarantees performance by parallel processing of information in the data warehouse. In
a traditional database, queries are simple and usually require searching in a small part of the
database; a typical query would require the search for orders received from a specific customer.
For queries that are repetitive or similar, DBAs tune the database so that queries can be
processed efficiently. Since queries for decision-support are not predictable, this is not possible
with data warehouses.
However, the processing capacity of a node is not permanently allocated for a particular task by
tuning it. Instead, Teradata has reserved the task of allocating disk and processing capacity to the
file system which does this logically and without any intervention from the DBA.
Currently, storage service providers determine their service level agreements based on not only
their disk capacity but caching as well. For the decision support that Teradata provides to its
customers, traffic is not predictable enough for caching to be relevant and cannot help to boost
performance.
Storage area networks are useful for load balancing required for high throughput traffic. However,
the throughput for CRM databases is not large but the numbers of I/O can be variable and large
although the volume of data for each of them is not unmanageably large.
The other potential benefit of storage area networks is the possibility of mirroring for convenient
access locally and to wireless devices. Teradata, however, centralizes the data at a single point
and in most cases publishes data to enterprise intranet in most cases. Wireless access to data is
possible but this task is left to middleware developers and to the suppliers of wireless devices.
Teradata has organized its data warehouse into nodes, each equipped with directly attached disk
arrays that are required to manage a data for a section of the table. So, a query may require the
search of a thousand rows across several tables. A typical case could be click stream data of
consumers which would include data on what they purchased when and from where. A query
would involve summary of data of all consumers at any location or time. This would be followed
by sorting to do analysis such as where among all the regions was a particular product more
popular than in other places. Each node is required to process data from a fraction of the rows
and return the results.
5.4 Location Based Services
Companies in the location based services space will gain an edge as they add attribute
information to their stock of spatial information. Most companies in this segment initially began by
providing simple services such as directions information and driving directions. Their current
plans will help them to differentiate themselves by adding services tailored to the customer group
they have acquired. MapInfo, for example, initially offered GIS information for infrastructure
planning services to clients in the telecommunications industry; in the future it will provide
information services for its field forces. Vicinity offered address and driving directions to local
outlets of retail chains and other multi-location companies and plans to grow by introducing
inventory information for the nearest store. Space Machine is adding services that enable field
forces to co-ordinate their activities as a group.
As the size of the databases increases over time especially with the addition of satellite imagery,
aerial photography, etc., the costs of delivery of information will be a critical determinant of
competitive advantage. Storage companies like SANZ (http://www.sanz-
inc.com/PDFs/ArchivalSystemsnocols.pdf), in partnership with ESRI, and Storagetek
(www.storagetek.co), in partnership with Space Imaging
(http://www.geoplace.com/gw/1999/0599/599data.asp) take advantage of HSM (Hierarchical
Storage Management), or a combination of disk and tape and optical disks in the future, to lower
the costs of management of large volumes of data. HSM uses a file system to automate the flow
of data from disks or tapes, from the media or the cache, by an orderly process that minimizes
the costs and latencies in data recovery.
Tape systems, traditionally, were reserved for archiving due to their higher latencies in data
retrieval. The delays include installation of the tape drive by a robotic arm, the search for the
required file and the time required for seeking the specific piece of data in the file. The time
required for the first task is reported to be 40 seconds, the second task takes 53 seconds, and the
seek time for the third task depends on the read rate of the tape55. The reduction of latencies
requires related data to be stored contiguously, organizing data for its parallel retrieval, optimal
use of caching by storing more frequently data on disks. File systems help to optimize the flow of
data.
SANZ has introduced the GEMS file system organizes information on tapes. An added bonus of a
file system is that data can be recovered in parallel streams once it is known where the
information resides. Finally, GEMS is a 64 bit file system which allows users to disperse the data
over several more storage devices and recover them in many more streams than is possible with
a 32 bit file system.
5.4.1 Space Machine (http://www.spacemachine.net/)
Space Machine is our choice of a company that is leveraging storage technologies to push the
envelope for location-based services especially in the enterprise space. Not only does Space
Machine go beyond using the staple Oracle database for storage of information, storage
technologies (SAN or NAS or RAID) are an integral component of the design of its architecture.
Data storage forms an independent tier in its 5-tier architecture that is an aid in achieving the
quality, price and speed of delivery parameters of its applications. The full impact of Space
Machine’s technological edge will be felt when the current bandwidth limitations of wireless
55 The background on Latency issues has been quoted from “Distributed and Hierarchical Storage Systems”, by Craig J Patten, K A Hawick, J F Hercus and A L Brown, January 1999, University of Adelaide, Australia
networks are overcome. Currently, Space Machine’s more advanced location based services are
available as web based applications accessed on the Internet.
Users of wireless Internet frequently experience delays in receiving data and in the presentation
of the information in an intelligible format. The separation of the application tier from the storage
tier in Space Machine’s architecture plays a critical role in lowering the delays in the presentation
of the information requested by the user. Map data and its depiction on a wireless device are a
common denominator in all location based services and this information changes very rarely if at
all. The rendering of map data, each time it is requested, is therefore wasteful of processing
power. By a simple expedient of caching the rendered map data on the server, Space Machine
makes it possible to receive this information almost instantaneously.
The data store plays an altogether different type of function of maintaining dynamic information
that is updated more frequently than map information. The data elements include traffic data,
street vectors, route data, address data and the list expands as the needs grow. When a user
requests, for example, information on landmarks in a neighborhood, the rendered map is
retrieved from the cache whilst the corresponding landmarks data is received from the data store.
Two major efficiencies are realized from relegating data storage to a storage area network or a
similar device. A server combining the caching, the application management and the storage
functions necessarily involves a high-powered processor, which is inevitably expensive. By
contrast, data storage is achieved with equal efficiency but at a much lower cost by cheaper
storage device.
Time delays in receiving information on a wireless device are accentuated when users execute
more complex queries. In the early days of location-based services, users could make-do with
routing information. Increasingly, they look for more of related information such as landmarks.
The data store is able to relay several types of information simultaneously much more efficiently
than a server would able to do.
Whilst the data store is a repository of all information, this does not imply that it is available at
only a single centralized location. Instead, copies of the data are mirrored at several locations
much like cell sites in metropolitan areas. The capacity for storing data is scaled as the number of
customers increases. The availability of copies of the data, close to the consumption point, helps
in lowering the time to download information on a wireless device.
Space Machine has made a deliberate choice not to outsource the function of data storage to
content providers. Component based architecture enabled by Java, XML/SOAP has encouraged
many players to specialize in application and business logic while the storage function as well as
content management is farmed out to partners. Space Machine has chosen to manage the
storage function to ensure performance and robustness. In its architecture, the application layer
knows exactly where it can receive its information.
Improved performance is also achieved by reserving the function of updating and of cleaning data
for the normalization layer. This tier in its architecture collates data from a variety of sources
within the enterprise and other content providers. Space Machine has written adapters that
prompt requests for updating information from data sources at stated intervals. The quality of data
is achieved by checking the consistency of the data with their common addresses.
The front-end business logic of the applications offered by Space Machine reflects the distinctive
flavor of its services. Individuals do not have to suffer the tedium of feeding their location
information. Instead, sensors built into their devices are able to identify the location. Servers, at
the back-end, push information to wireless devices. Proximity-based triggers that are written to
user specifications drive the information to an individual’s device. Each user receives information
that is relevant to his or her location.
The user specifications are stored on the data store tier. When wireless devices are lost, the user
specifications are easily restored. In Space Machine’s experience, the ability to preserve user
specifications is an important consideration for carriers because it helps to retain customer loyalty
This specific design is suited for users whose location affects decision-making. In the enterprise
space, field forces have a use for Space Machine’s applications but not sales staff predisposed to
CRM applications. The whereabouts of members of a field force affects the allocation of tasks,
time allocation and information about the jobs that they have to complete.
In general, Space Machine has learnt that adoption rates for enterprise applications are faster
when they are tailored for specific work-groups instead of for the entire firm. Once acquired, the
churn rate in the enterprise space is much lower than the consumer space.
In the consumer segment, the same technology (location aware) can be used to keep in touch
with the members of the family, friends and pets. Thus, parents can be alerted when their young
children go beyond neighborhood boundaries. Similarly, users would know when their friends are
in the vicinity. Finally, owners of pets can monitor them.
Space Machine does not feel it’s hamstrung by the inaccuracies in the information received from
position identification technologies currently available. GPS data is error prone because of the
delays in receiving information. This can be corrected by software that adjusts for the predicted
errors in the information. The deficiencies in the information on cell-id is corrected by consistency
checks of inter-related information such as the number of blocks separating two locations helps to
correct for any errors in distance information received from cell-id data.
At this point of time, Space Machine’s data needs are in the terabyte range but not all of it is used
by wireless applications. A great deal of the space is required for aerial and satellite imaging that
is delivered over the Internet and are required by a single user. Space Machine has the ability to
offer raster data but is constrained by bandwidth availability.
In the near term future, data storage needs will grow as both the enterprise and consumer market
are offered more detailed information with related routing data. For example, customers will be
able to identify not only restaurants in the vicinity but also the menus offered. Enterprise
customers will be able to track their assets in their locations.
The applications developed for the web enables Space Machine to support a data store as well
as to prepare for richer applications that can be offered, in the future, for wireless devices. As
bandwidth increases, web based applications will also be available to wireless customers.
5.5 Medical Applications
Health enterprises had invested in IT for only peripheral activities such as imaging on the
one hand and administrative processes on the other. Clinical work flows, the heart of the
health system, was largely untouched.
Storage companies that are able to integrate clinical work flows will gain competitive
advantage. Automation of clinical work flows and associated decision support advantages
would provide an edge to the health enterprise and to the storage companies that are able
to best able to provide these solutions.
The challenge of automation of clinical work flows is their diversity and the overriding role
that human intervention plays in them. Consequently, the vocabulary and the associated
data definitions are contentious and diverge between enterprises. The winning companies
will be those who are able to offer acceptable data definitions and lay the basis of
automation of clinical work flows.
5.5.1 General Electric Medical Information Systems
http://www.gemedicalsystems.com/it_solutions/index.html)
GE Medical Information Systems (GEMS) is our choice of a company that is leveraging storage
technologies to launch disruptive applications in the health industry, which includes wireless
applications for data capture. Unlike its peers in the industry, GEMS is extending its applications
beyond medical imagery into clinical information systems that have been largely resistant to
automation. GEMs bid to integrate medical imagery with clinical information systems open the
way for a variety of decision-support applications in the health industry. The distinctive flavor of
the design of its information systems is that they revolve around the needs of clinical
professionals rather than administrative or transactional workflows in the health industry.
GEMS has reached an inflection point in its growth where it is able to integrate multiple data
types like medical images from radiology, wave forms from cardiology, clinical data and vital signs
data to be able to create the infrastructure for centralized management of information received
from several sources within a hospital. The information will be aggregated such that the paper
shuffle between departments within a hospital or a medical group will be minimized as well as
between individual units of a hospital complex or medical group.
The turn in its business strategy is reflected in the spin-off of the information systems business
from GE Medical Systems, several acquisitions, and its recently announced partnership with
Qwest, which will manage its storage and related software at its data centers.
GEMS partnership with Qwest marks a new era in its corporate strategy. At its data center in
Chicago, Qwest will host the storage sub-systems supplied to it by EMC. The hosted application
services will connect to hospitals, within a group and across regions, by fiber optic networks
including on the last mile. The speed of operations and the reliability of the network will be critical
to persuading an increasing number of customers to subscribe to the hosted services.
GEMS existing customers use an intranet and an archive is located off-site on a data center. In
the initial stages, radiology was the focus of digitization of images and storage as the volumes ran
into ten to fifteen terabytes of data. The images, together with the notes from the radiologist (the
Radiological Information System (RIS)) are archived off-site in a data center. The Picture and
Communications System (PACS), for retrieval of images, resides on premises due to insufficient
bandwidth.
In the near tem future, both PACS and the archive of will move off-site. As a result, it will be
possible to have a single system for archiving of pictures as well as for communication to multiple
departments. Thereafter, data from the cardiology department will also be centralized at the same
data center. In the more distant future, patient data and other images will be made available at a
single data center. The aggregation of multiple functions will pave the way for launch of several
applications.
Hitherto, GEMS customers were small and medium sized hospitals who were constrained by
space and resources and recourse to hosted services is a means to keep pace with technological
developments. Large research hospitals had the resources to invest in in-house IT and were
unwilling to risk any loss or disclosure of information by outsourcing. The speed, reliability and
size of operations substantially improve the benefits/cost ratio enough to persuade the larger
hospitals to consider outsourcing. GE has also developed advanced security software to raise the
confidence level of its customers.
The process of outsourcing of applications has begun with clients who subscribe to its online
services. In the year 2002, an accelerated pace of implementation will happen since 50
customers have entered into contracts and altogether an additional 100-150 customers are
expected to subscribe to its services. As services are outsourced, the current rate of growth of IT
related revenue, at 10-15%, can be sustained if not exceeded.
Workflows involving patient information are harder to automate as health data is viewed in a
context and from the lens of a particular discipline. IT professionals have also grapple with
idiosyncratic trade practices of physicians, and the professional codes of doctors and nurses. In
the past, projects of this nature were sponsored by the National Institute of Health and
spearheaded by Chief Medical Officers who were oblivious to the needs of the practicing
professionals. Consequently, the success rate over a thirty-year period has been low.
GEMS strategy will chart a new course and will be based on the premise that the priorities of
clinicians should be coordinates of the systems that automate the workflows in a health system.
This would imply that workflows would be sub-divided in a way that templates of information
would be customized for each type of professional. The other important assumption of its strategy
is that the health enterprise is fragmented at several levels and the value of automation is not
equally compelling at each stage. That the adoption rates will be the highest if the most
dysfunctional work processes are automated first.
The first stage in treatment of patients, the review of their data, is a common denominator in all
health workflows and the most valuable to doctors and nurses. They all need background
information on a patient to prepare for diagnosis and prescription or undertake any surgical task.
A clinical repository, with records of longitudinal patient information, medical history, laboratory
results and encounters with physicians, will form a central repository accessible to any doctor or
nurse. A clinical repository is, therefore, the bedrock of any attempt at automation. The paper
version of a patient chart entails considerable opportunity costs, monetary costs of $40 per event
besides time lags and incomplete data.
The complexity of automation is the greatest at the stage of patient care or clinical encounter
since workflows are specialized and heterogeneous with differing priorities for patient safety,
quality of care, productivity and operating efficiency. Its here that doctors make a judgment call in
individual cases or assessments varies depending on the wealth of information available and how
promptly it’s delivered.
The nature of workflows is dissimilar in an Emergency Room, an Operating Room, the Intensive
Care Unit, Labor and Delivery, the wards and the outpatient clinics. In the operating unit, time
delays are extremely costly since they utilize capital-intensive equipment; the estimated
opportunity cost is estimated at 19 cents a minute for each second that lapses. The delays occur
because patient consent and signatures are necessary, Anesthesiologist’s report is held up, etc.,
which keeps doctors waiting. By contrast, operating efficiency is not as much of a concern as
quality of care in the Emergency Room; doctors and nurses need the white board with information
on patient’s conditions and needs, any work conducted prior to entry into the emergency room
such as laboratory results and reports.
GEMS strategy for automation of specialized domains of medical care is to develop a care plan
for each of them and the corresponding workflow schema for each of them. The functions of
doctors and nurses are delineated and the information needs to reduce their inefficiencies are
identified. GEMS will create templates for each stage of patient care and a clinical event manager
will prompt the users when a task is required to be completed.
GEMS expects that a great deal of the resistance to clinical information systems will ease when it
gathers data about patients and aggregates it in a data warehouse for decision support. A
growing wealth of readily available knowledge takes away the strain of remembering
pharmacological and other information and the errors that occur when doctors forget them.
Consequently, decision support information systems are readily accepted.
The patterns of disease change over time as well as the state-of-art of diagnosis and cure.
Today, for example, anthrax has emerged as a completely unforeseen health risk. In not too
distant a future, biotechnology will spawn a new generation of medical practices. GEMS have
created templates that can be customized by health professionals by providing a pick list of items
that can be added to the template. Furthermore, the pick list can be changed by consensus.
The third and final stage of automation is the processing phase or tasks such as order entry,
charge capture and post-care consultation. At this stage, errors are common because
pharmacists often don’t understand the physician’s handwriting, drug interactions are not taken
into account or susceptibility to allergies is not known. At this stage, acceptance of automation is
greater because the value proposition is manifest.
6 EXPERIENCES OF KEY PLAYERS:
6.1 Storability (www.storability)
Storability is a storage service provider, which offers storage management services with a slant
towards service quality. It builds on software available from Veritas, for centralized management
of storage area networks, to automate processes in order to provision, monitor and report on
performance expected under service level agreements. The software that it adds for ensuring
service quality provides it the edge in the marketplace. Storability’s market is very largely the
security conscious enterprise market, which expects services to be offered behind the firewall.
Over the last six months, it has entered the telecommunications space where the large TELCOs
have garnered business from enterprise customers.
In the enterprise segment of the marketplace, Storability has witnessed a trend towards
increasing consolidation of storage assets within an enterprise. Increasingly, companies are
looking to integrate the islands of storage devices scattered across several departments into
storage area networks. In the past, each department purchased storage devices randomly
ignoring available resources elsewhere in the enterprise.
Besides reducing costs, centralized management of all storage assets creates an independent
business within the enterprise, which offers services to its internal clients. The internal clients
expect services to be customized for their needs. Storability’s services are useful to meet the
contractual agreements between internal clients in the company.
In the telecommunications space, Storability made its entry by partnering with CTC
Communications, one of the few CLECs that have survived the current downturn. In addition,
Storability is awaiting the conclusions with over six other telecommunications companies
including Qwest, which has been most successful in obtaining very large deals of over 100 TB.
The clients in the telecommunications companies are looking to offer especially content
management services and e-business services to their customers. Customers are more willing to
trust well-established TELCOs to outsource storage services. Storability’s services help the
TELCOs to meet the technological requirements of service level agreements.
The demand for content management services has transformed the competitive dynamics in
Storability’s marketplace. Companies need a diversity of services, at varying levels of quality and
price, for rich media management, live video, medical imaging, besides storage of e-mail required
to observe FCC’s regulatory requirements.
Storage management services have to be tailored to meet the requirements of content sensitive
applications so that it is not enough to have generic databases. For example, rich media makes
greater demands for scalability and is harder to manage so it costs more. Other storage services
include content filtering.
Each of these verticals attracts start-ups who specialize in one or the other niche markets.
Storability, in this situation, sees an advantage in licensing its software as a platform, which
others can use to offer customized services. Additional revenues can be earned from the
certification of licensees who have to meet acceptable levels of quality before they can get to use
the platform. Curiously, Storability has received requests from customers to certify their own
internal clients be certified before they use the platform.
6.2 IBM/Tivoli
IBM, together with its partners, offers the entire gamut of services in the value chain of storage-
enabled applications. The services include storage sub-systems, software for storage space
management (volume and virtualization software), content management, storage data flow
management software, web applications servers and web services including wireless
applications. These complimentary capabilities enable IBM to assemble the pieces of the systems
that will transition pilot projects, demonstrating concepts of storage-enabled applications, into
industrial scale operations.
Several pre-requisites have to be met for the implementation of IBM’s strategy for industrial scale
applications. These include streamlined processes; higher speed of retrieval of information,
automation of content management, pre-determined standards for quality of service, and data
flow management. IBM’s Content Manager, Tivoli Storage Manager and storage products like
Enterprise Storage Server are the tools intended to achieve these objectives.
Beyond workgroups, the challenge of content management is to distribute slivers of data from
centralized repositories to points of consumption. Also, the data has to be also transformed to
adapt to the format required by the client. Caching has relied on statistical probabilities to store
data locally largely for desktop web browser clients. Statistical probabilities are harder to estimate
for nomadic wireless clients. Consequently, content delivery networks would have to replicate
much more data to meet consumer need unless alternative intelligent software management
solutions are able to distribute data in real time. Content management continues to evolve within
the industry and IBM has acquired Open Markets (www.openmarkets.com) and Ascential
Software (www.ascentialsoftware.com), even though it has its own Content Manager, to
strengthen its capabilities.
The IBM Content Manager was initially deployed for still photography applications such as the
Vatican Library, The Hermitage and the Andrew Wyeth collection. It was then upgraded to
manage video collections. Customers have been wary about adopting video content management
tools, especially the more complex applications, since they don’t yet have experience with the
system integration requirements.
Tivoli’s software products, Tivoli Storage Manager (TSM), Tivoli Storage Network Manager
(TSNM) and Tivoli SANergy, are meant to streamline the flow of data from data storehouses to
applications in an environment of heterogeneous operating systems and applications. TSM, the
flagship product, is a suite of modules focused on back-up and recovery. TSNM provides SAN
management functions including automatic discovery, topology mapping, and policy-based
capacity-on-demand. SANergy provides a logical view of resources on the storage area network
and manages data sharing across nine different operating platforms.
Tivoli is adding new functions to its products to make them capable for industrial scale operations.
The latest version of TSM incorporates LAN-free back-up, tape sharing, and enhanced mobile
computing backup capability. The full utility of Tivoli’s software will not be realized until there is
greater file mobility between storage devices and servers and vice versa. For example, a program
operating on a Windows NT platform cannot access a file on a UNIX server; it has to be available
on a file server or has to be redirected from the latter to the former. Full deployment of Tivoli’s
products awaits the APIs that will facilitate inter-operability.
One of IBM’s storage offerings, the Enterprise Storage Servers, ( code named Shark), is
designed to meet the demanding response time requirements of applications, such as rich media,
by combining advanced disk drives, cache, advanced algorithms. The mechanical properties of
disks limit their ability to meet response time requirements for applications like rich media. Shark
increases response time by using intelligent, adaptive algorithms to change the caching logic to
adjust to changes in I/O access characteristics in real time, and increases disk system throughput
by accessing multiple disk drives in parallel56.
Having acquired the technical ability to work in an open SAN environment, IBM is pushing the
industry to adopt standards for inter-operability. Its SAN exploitation software works on a variety
of platforms including zSeries, mid-range iSeries and UNIX/NT. IBM’s SAN exploitation
applications (TSM, Tivoli SANergy and TSNM) can work with AIX, Solaris, HP-UX, Compaq’s
Tru64, Windows NT and 2000, Novell and Linux and the mainframe works with ESCON.It has
taken the initiative to promote open standards for SANs in forums such as SNIA, FCIA, IETF,
ANSI T11, DMTF and ISO among others.
Meanwhile, it has chosen to pursue the commercialization of rich media applications, require
management of blocks of data and do not need to use heterogeneous operating systems.
Currently, IBM is limited to using the AS 390, its own operating system, for the deployment of
SAN enabled applications
For the high end of rich media applications for large broadcasting companies, IBM offers a file
system called GPFS (General Parallel File System), which works best in conjunction with its 56 The details of the architecture can be found at http://www.storage.ibm.com/hardsoft/products/ess/essperf.pdf
supercomputers but can also be used with lower performance servers. The key feature of the
GPFS is its ability to manage numerous video streams running in parallel; blocks of data are
broken into stripes for read and write operations on several disks.
Like other virtualization software, GPFS provides a logical view of the physical disk capacity and
allocates data using a volume manager. Its distinctive character is that it manages parallel
streams of traffic. A cluster of file nodes, each of them having a complete view of disk capacity,
access a corresponding array of disk nodes intermediated by a switch.
GPFS file systems ensure very high levels of performance since they do away with file sharing.
Non-linear editing happens with the entire team sharing the same storage, which can reach
multiple terabytes of data. The speed of operation can reach 100 Gigabytes per node.
GPFS was used in early video-on-demand trials in the mid-nineties but floundered when it could
not be run on modest bandwidth then available. The needs of cable over builders, for their
interactive television programming, and encoding needs of large broadcasters have revived the
prospects for GPFS.
Interest in broadband distribution projects has begun but the pace of progress is constrained by
previous investments in analog systems. Due to these legacy investments, the large majority of
broadcasters is unwilling to switch to digital storage systems and associated software except for
editing functions. Currently, IBM, in partnership with e-motion, is working with a leading British
broadcaster to archive its footage in digital form.
However, none of the customers have shown interest in GPFS and prefer entry level, lower end
storage systems57 that IBM has to offer. Einstein TV, www.einstein.tv/uk/index.asp, a British
company that broadcasts science documentaries, was among the first to use storage for digital
content management. Einstein TV deployed IBM’s entry-level systems (TV-in-a-box) that can be
installed within twenty days.
Adoption of IBM’s entry-level storage infrastructure systems is more extensive and now includes
the police departments in Great Britain. Video surveillance cameras track the movement of cars
in parking lots of crime-prone neighborhoods. The volume of visual records with police 57 Links to a detailed technical description of the systems can be found at http://www.storage.ibm.com/press/announce/20010921.html
departments is growing as digital cameras are installed in squad cars as well. IBM’s systems are
used to store, retrieve and analyze the information gathered with these cameras.
Dalet Digital Media Systems, a French content management company, has used the same
systems for satellite audio (XM radio) installed on luxury cars supplied by General Motors.
Storage needs of XM radio are high because it offers 100 or more channels of music.
IBM’s tape systems58, for rich media, is the other product that has gained market acceptance.
Entertainment Tonight uses it to cut costs of ingesting and logs of video tapes.
The rate of adoption of IBM’s SAN solutions is expected to increase with the expanded scope of
Einstien.tv’s future plans. By 2002, Einstein.tv plans to broadcast its content to sixteen countries
in six languages, which would necessitate conversions to several different TV standards,
translation into several different languages besides adapting to a variety of devices. Furthermore,
footage will be received from several different sources. A centralized repository, together with
content management tools, enables Einstein to lower the costs of conversions and distribution of
content to several different points of consumption.
6.3 Veritas (http://www.veritas.com/)
Veritas, the leader in storage management software space, offers an application NetBackup
Professional for mobile professionals. The product automatically saves corporate data that has
not been previously saved on laptops and mobile devices onto a central repository and can
recover data when it is lost. NetBackup( http://eval.veritas.com/downloads/pro/NBU-
P_white_paper_v3.pdf ) addresses the need to protect corporate data that is frequently lost when
laptops and other mobile devices are either damaged or stolen when executives are traveling.
The data on these devices is often of paramount importance to a company since mobile
professionals typically play key decision-making roles.
The actual adoption of the NetBackup product has not lived up to its promise despite the
opportunity costs of lost data, which can be as high as a quarter of a million dollars in some
cases. Customers continue to prefer synchronization with their desktops, which are connected to
58 Technical description of tape systems can be found at http://www.storage.ibm.com/hardsoft/tape/3584/index.html
a central repository. Price resistance is the chief reason for the cold response from the
customers; ASPs offered the product at a monthly charge of $ 10-15 for each user whereas
consumers are not willing to pay more than $ 3-5 a month for each user (even when the charge is
paid by an enterprise). At the peak of the boom in the year 2000, the ASPs were overly confident
that customers would be willing to pay their asking price.
The mood in the industry has changed following the slowdown in the economy. ASPs are
experimenting with vertical product differentiation, i.e., a broader range of prices and quality of
service for different segments of the consumers. They are realizing that customers have different
perceptions of urgency of recovery of data; price levels have to be adjusted to reflect the
expected level of service by individual classes of customers. The jury is, however, out on the
eventual success of the emerging pricing policy.
6.4 E-motion (www.emotion.com)
E-motion, a Los Angeles based media management software company, typifies a player that
takes advantage of storage technology to develop applications for a niche market in media
production management in the studios and the advertising industry. It develops software that
currently aids editing functions. The company is not convinced that a business case exists for use
of rich media content in other applications such as customer relationship management or supply
chain management.
E-motion caters to the enterprise market because its software tools, eMedia Partner
(http://www.emotion.com/mediafiles/pdf/mp_architecture.pdf) and Global Brand Manager,
are an aid to manage work processes in editing processes including storage and search of very
large repositories of rich media assets such as collections of photographs, corporate intranet,
video libraries within the media industry. Its software tools are used for archiving of digital assets
as well as for their distribution including e-commerce in digital assets. A distinctive feature of the
tools is their ability to search a piece of media by natural language keywords.
A typical customer of E-motion is an advertising company that needs to keeps its clients informed
so that a campaign can be organized within the scheduled time. The storage of work-in-progress
saves the time and tedium of waiting for customers to approve prototypes.
The preference for the enterprise market is reflected in E-motion’s recent expansion in Europe
instead of Japan. It acquired its reseller in the UK, Integrated Solutions Technology (IST), which
will also manage its business in continental Europe. Although E-motion’s partners have
encouraged its growth in Japan, E-motion is not geared to meet the demands of its growing
consumer markets for especially wireless applications.
In the near term future, E-motion sees a business case for extension of wireless capabilities, in
the enterprise market, for its tools especially in the European market and its new acquisition will
cultivate this market. In the USA itself, the market is small and restricted largely to collaboration
within the movie studios. One of its largest customers is BBC, which is looking to reduce costs of
conserving its footage.
The value proposition of E-motion’s software tools is the economy realized by geographical
dispersion of individual components of work processes in media creation to regions that excel in
them. For example, the shooting of footage can take place in East Asia, its processing may best
be done in the USA and it may be broadcast in Europe. However, scattered production can be in
conflict with time schedules when means are not available to quickly co-ordinate the activities of
works groups. Storage technologies facilitate collaboration by mirroring copies of work-in-
progress across geographical regions. Wireless capabilities are another aid to meet time
schedules by allowing instantaneous access for mark-up and approval of work completed.
E-motion’s products are sold as licenses to large enterprises or smaller companies pay a monthly
fee its ASPs, PictureQwest and Footage.com. As a software company, E.motion does not see
its long-term interests in the ASP business and recently sold PictureQwest to Creatas LLC.
6.5 Viafone (www.viafone.com)
ViaFone is a mobile applications platform (going by the moniker ViaFone Bridge) provider that
currently focuses on services deployed from behind the firewall. ViaFone has designed its servers
to offer both data and voice information services, with speech recognition technologies from
Nuance, for mobile phones and data devices. ViaFone expects its strategy to be more effective
as devices that can receive voice and data become available in the near future.
After testing the waters in the consumer market, ViaFone repositioned itself for the enterprise
market. The change was necessary to integrate the Customer Relationship Management and
Supply Chain Management databases of customers with user interfaces for applications that
ViaFone develops for salespeople and field forces especially in the electronics and
pharmaceutical industry.
For the consumer market, ViaFone continues to provide hosting services required to deliver
MySimon’s comparison-shopping web content to carriers including AT&T, Sprint and Verizon.
The experience prepared ViaFone to tailor content for a variety of devices used by customers, to
scale up its operations and to sustain the service for over a year.
The combination of voice and data services is especially useful for its vertical market in the semi-
conductor industry where field staff has to work from inside dust-free zones where the use of
mobile data devices is prohibited. Field maintenance staff had to bear with the arduous process
of moving in an out of these regulated zones, besides taking-off their hoods and putting them
back, before they could use their mobile data devices. The access to voice information services,
received on phones, saves all this trouble.
In addition, voice information services can reach locations where wireless networks don’t have
coverage. Typically, wireless signals don’t reach basements. Salespeople in the pharmaceutical
industry travel a great deal and can use their time to receive information while they are driving.
The advantage of voice services is gained at relatively low incremental costs of 5 to 10% in
situations where customers already have databases.
ViaFone is required to cater to enterprises’ need for security and integration with a variety of
databases and protocols, including the legacy versions, to be able to work behind a firewall. In an
outsourcing situation, on the other hand, the environment is not as diverse since enterprises do
not farm out mostly web content and keep mission critical information services in-house.
The services offered by ViaFone meet CRM and SCM information needs of salespeople and field
engineers. Salespeople require information on customer demand and order flow, the staple of
CRM databases, and also make use of remote file management services such as faxing
documents on customer sites. This is especially true in the pharmaceutical industry where a great
deal of product information is exchanged. Field staff alone use the supply chain management
databases to manage deliveries from manufacturing to be able to meet delivery schedules. They
also use SCM databases to communicate information from customers, in real time, to ensure that
equipment manufactured for them is customized for their needs.
Tellme (www.tellme.com)
Tellme integrates speech recognition technology from Naunce with back-end Internet
technologies to provide voice information and application services. It has two types of markets—
the consumer market which it serves in collaboration with AT&T Wireless. The other market is the
enterprise market where it caters to several verticals.
In the initial stages, the quality of voice recognition had to be substantially improved especially
because consumers’ queries are not the standard variety. Prolonged testing has tripled the levels
of self-service.
Whereas traditional IVR technology used proprietary technology, Tellme uses VoiceXML, to take
advantage of component based web services model to integrate with the standards based
Internet Infrastructure. Unlike IVR, VoiceXML does not require a separate infrastructure for voice
information services.
By separating the location of the web application logic from data storage, the Tellme is able to
take advantage of scale economies possible from outsourced storage services. The back-end
infrastructure is managed by Exodus Communications. In addition, Tellme is able to allocate
storage on demand in sync with seasonal fluctuations of call volume in individual industries.
In a typical web services strategic model, Tellme offers an application development platform with
a library of components that developers can use to design customized solutions for their
customers in individual industries. Tellme does the grunt work of integrating computer-telephony
equipment with data centers, upgrading speech recognition software to respond to varying
accents and to manage the network while the developers have an operating infrastructure to run
their applications.
For the enterprise market, voice recognition technology is used to automate call center functions
and to provide notifications. Typically, the automation of call center functions includes routine
information such as flight information for airlines, location and driving instructions for banks, stock
quotes and company information for brokerage houses, etc. Notifications are used to prompt
users about say scheduled flights.
6.6 Openwave (http://www.openwave.com)
Openwave entered the market with software solutions for wireless applications, such as unified
messaging, based on the WAP protocol. It has expanded the portfolio to include VoiceXML based
products and has plans for I-mode based applications. It plans to offers its voice applications,
based on VoiceXML, to carriers in partnership with CTI2, CISCO and I-Basis. CTI2 contributes its
WW Messenger, a unified mail-box that receives and responds to messages from phone, fax,
telephone or a wireless device. I-Basis is an ASP that hosts the open systems unified messaging
solutions on behalf of carriers and Openwave does the backend integration. Openwave integrates
CISCO’s voice gateways with VoiceXML to offer voice portal services.
Openwave switched from EMC and Sun Microsystems for its storage needs to Network
Appliance, a company that specializes in network attached storage. The original suppliers
insisted on retaining their proprietary systems while Network Appliances was responsive to open
systems that would collaborate with other competing vendors. Openwave reported much lower
cost of installation of storage devices from Network Appliance besides ease of migration to
competing storage products when they are needed.
Despite the ease of implementation of proprietary systems provided by EMC and Sun
Microsystems, carriers found that their costs of storage management would rise as they added
new applications such as unified messaging. Each time they add an application, they had to
scrap their previous investments in hardware and software. Furthermore, storage costs are rising,
as much larger voice files have to be accommodated in the devices. Network Appliances
provided better value for money.
6.7 Datadirectnet ( http://www.datadirectnet.com/)
DataDirect Network is focused on providing storage solutions for a range of rich media
applications and video-on-demand is the thrust of its current activity. It builds storage appliances
that combine the functions of RAID controllers, Fiber channel switches and a cache to reduce
latencies in data transfers for streaming media applications. It has partnerships with InfoValue
and Vsoft to provide video streaming solutions.
The SAN Data Director, a SAN appliance and the key product of the company, has been
designed to provide the efficiencies required to achieve the quality of service that is a prerequisite
for rich media applications. Latencies in data flows from the storage device to the streaming
server are reduced in the following ways
Fiber channel switches and RAID controllers have traditionally mediated access to
storage disks. Latency penalties are incurred as data flows from each of these RAIDs to the Fiber
Channel switch. The delays cascade when multiple layers of switches are required to transfer
data in larger systems. SAN Data Director eliminates these delays because it has twenty
embedded Fiber Channel ports that receive data in parallel streams. When several servers call
data from storage devices, they can make conflicting demands on switches and shared RAID
controllers, which lowers the throughput of data. By contrast, data flowing in parallel streams
does not cause such problems.
SAN Data Director adjusts the size of the data block to the volume of data flow.
Traditionally a block of data is broken down into stripes, which are distributed over eight disks.
The unit cost of processing a stripe of data rises as the size of the data block decreases. SAN
DataDirector uses a single LUN with varying numbers of tiers; when volume of data rises the tiers
increase and vice versa.
Most of the data transfer to the host takes place from a cache at electronic speeds, which
are higher than data transferred from physical disks. High Speed Traffic Directors(HSTD), a
software built into the SAN DataDirector, enables the transfer from the cache to the host. Delays
in searching data from the cache are minimized in several ways; the HSTD manages the cache,
aided by the company’s proprietary software, such that the least used pages are discarded. In
addition, another piece of software tracks related blocks of data and schedules their consecutive
transfer when any one of them is required. Finally, proprietary software developed by DataDirect
Network searches data from the cache in real time before the connection with the channel is lost.
Efficiencies in transfer of streaming media is achieved by sending a single master copy
from one end to another instead of congesting the infrastructure with numerous streams that
enter the network when numerous viewers watch a program.
For wireless customers, data is transmitted in eight parallel streams to avoid the
congestion that will follow when data has to be transmitted to several towers in a region.
6.8 Generic Media (www.genericmedia.com)
Streaming media has not been widely adopted because of the high cost of its delivery and its
inconsistent quality. For media companies, on the other hand, streaming media is potentially an
attractive alternative channel for distribution that can expand the market for their content.
Streaming media meets distinctive needs since it can be viewed on demand and at users’ site of
choice. Generic Media is one among several other companies that have set out to reduce the
cost of delivery of streaming media as well as to improve the users experience of viewing.
One of the sources of inefficiency in the delivery of streaming media is the cost of producing a
large number of copies of content each of them tailored to a particular media player such as
QuickTime. In addition, the bit rates for each of these players can vary to suit the bandwidth
available with a user. Furthermore, media companies have to cope with the uncertainty that is
created by the emergence of new media players and changing availability of bandwidth.
Generic Media’s Media Publishing Service is a means by which a master copy of the content can
be created that can be transformed for any other format and bit rate automatically. The software
is constituted of three components—the Generic Media Transformation Engine, the Generic
Media Publishing Manager and the Generic Media Delivery Manager. The Generic Media
Delivery Manager detects the users’ settings including the player, its format and the bit rates, the
Transformation Engine reformats the master copy to suit the environment and the Publishing
Manager configures the content for a particular user.
Such a service does not intimidate the user who is not required to make any selection of the
media media player the bandwidth settings or any other plug-in. Content can be downloaded from
the Internet by simply clicking the hyperlink.
Generic Media has also developed a bandwidth limiter that manages overflows of traffic that are a
common experience with streaming media. Access to a service is restricted before demand
exceeds the volumes that the pipe can channelize.
Finally, it has developed a gMovie Player for Palm Pilot for streaming media applications for
wireless devices.
6.9 Digital Fountain (http://www.digitalfountain.com)
Streaming media can mimic the economics of traditional media if its unit costs fall with the
expansion of its user base. The fixed costs in traditional media are incurred at the outset and unit
costs fall with the increase in the numbers of viewers. By contrast, the costs of delivery rise
linearly with streaming media as the number of users increase. This is because each client
requires a server each to complete a request for content and the bandwidth needs rise
proportionately. The math for streaming media will be more like that of traditional media if a
single master copy is delivered from the core to any edge site. Digital Fountain’s products,
Streaming Fountain and Download Fountain, find solutions to prevent costs from rising in
proportion to the number of users.
The traditional means of transferring data on the Internet, TCP/IP, was designed for short bursts
of traffic. It organizes data in packets that are sent across sequentially. In a broadcast situation, a
stream of packets carries small bits of information (payload) from a server to a client. The
complexity of carrying data in this fashion grows when packets are resent to correct for errors.
Digital Fountain’s family of products needs only a single server to transmit data to multiple clients.
The original content is transformed into metacontent or a master copy that is transmitted from a
server to numerous clients which could be an equivalent of a TV audience. Each copy of the
metacontent is composed of several components that are related to each other just as individual
variables are in a linear equation. The individual components are transmitted separately on a
network and reconstructed by the client. Unlike the TCP/IP method of transmission, the client
does not need all the components of the content to reconstruct the master copy. Instead, a
subset of the components is adequate while the rest are pieced together based on their inter-
relationships.
Digital Fountain’s technologies are meant for transportation that caters to the need for on-demand
viewing. The applications include training and corporate communications needs of large
enterprises who have to reach their employees in several branch offices, satellite
communications, video-on-demand (when a fully IP cable broadband access is available) and
transfer of data from the core to edge sites as well as data transfer on wireless networks.
Currently, Digital Fountain is implementing its technology in the satellite communications sector.
Typically, satellite communications companies transmit data three times from their Transponders
to earth stations to compensate for possible losses caused by atmospheric disturbances. This
would not be necessary with technology available from Digital Fountain since the information can
be reconstructed from some components that constitute the master copy.
In addition, Digital Fountain is negotiating with wireless carriers who have multiple subscribers
sharing bandwidth. The noise level increases as the numbers of customers and volumes of data
transferred for each user rises. Digital Fountain is positioning its products for attenuation of the
signal loss in transmission.
6.10 Airborne Entertainment (http://www.airborne-e.com)
Airborne Entertainment is one company that seeks to make the much-anticipated transition from
staple wireless applications, such as e-mail and calendars, to sale of entertainment content. Its
business model explicitly emulates NTT DoComo in North America in that it sees mass adoption
of wireless Internet access on the strength of content that consumers will be willing to pay for.
Airborne Entertainment specializes in short attention span entertainment, such as comic strips,
humor, etc., that can make-do with the available bandwidth with 2 or 2.5 G wireless phones. The
early success in winning business from the largest carriers, AT&T, Verizon and Sprint, points to
the potential for creating business for wireless content that will in turn create a demand for
storage.
The key planks of the business model that Airborne Entertainment is implementing are as follows:
Its content strategy excludes video content that it sees as unsuitable for the small size of
wireless devices in the near-term future as well as in the longer-term.
The distinctive value of wireless entertainment lies in its interactivity; FM radio provides
similar content but does not encourage the same level of participation from the audience.
Carriers increasingly see themselves as retailers of content that is the primary source of
revenue; advertisements, as the sole source of revenue, cannot be a viable model of business.
Publishing companies, like Airborne Entertainment, are valuable to carriers since they become
the conduits for sale of content produced by numerous small companies. In addition, they
provide customer support, service level agreements, etc.
Unlike the traditional media companies, Airborne has an edge in content distribution
technologies. Airborne’s proprietary technology, RUNWAY, can be tailored to meet the specific
needs of wireless carriers such as optimizing content for several different devices,
personalization of content, supporting SMS for a variety of protocols, billing, etc.
6.11 I-DRIVE (http://www.idrive.com/)
I-drive is a middleware company whose software provides solutions primarily for personal
storage requirements. Its products facilitate the storage, file sharing and file management of
especially rich media content. Typically, customers use its software to download documents from
the web, music files or store digital photographs that can be shared with family and friends.
Enterprise customers use its storage for saving documents that they can access remotely.
Till recently, I-drive offered 50MB of free storage where customers could download their files. The
software and the storage services were targeted at mobile users who could share files by
transferring them between accounts on the storage device, download from the web for later
viewing or print files without straining the limited bandwidth available with their mobile devices.
The middleware facilitates the interface with a variety of devices, transcodes content for a diverse
range of players and bit rates and mediates between varying protocols used for messaging.
The main source of revenue for I-drive in the early days was from its contracts signed with
twenty-five universities. Storage provided to students on campuses was paid for by the
administration from the tuition fees. Usage patterns in the universities indicated a strong interest
in multi-media content. Students used 55% of their space for multi-media files equally divided
between audio and image files while 40% was used for documents and 5 % for zip files.
I-drive has abandoned its efforts to cultivate a retail market for its software, which it initially
promoted by offering free storage. The company admitted that the free service was made
possible by venture capital, which could not any longer be sustained. As a middleware company,
I-drive does not see itself servicing the retail market where brands count. Instead, it is
collaborating with carriers and ISPs to offer its software as a package of value-added services
that consumers would be willing to pay for.
6.12 Deep Bridge (http://www.deepbridge.com/)
Deep Bridge specializes in developing content management solutions for text and pictures
repositories that are required in the pharmaceutical industry for regulatory purposes and in the
financial industry for regulatory reporting. The company has a conservative view of the
possibilities of content management solutions for rich media repositories; it sees a limited
potential for them in niches like mass media houses.
Currently, Deep Bridge does not see a prospect of remote searches of data on storage area
networks. The Application Programming Interfaces (APIs) required to search for information
managed by a variety of operating systems have not yet been developed. Similarly, efforts at
personalization, delivery of content independent of the media players, etc, have achieved limited
results on account of the slow progress in standards based applications development.
Deep Bridge does not foresee a significant potential for growth of markets since content
distribution to retail markets is hamstrung on account of bandwidth limitations. The modest
diffusion of broadband in the last mile and bandwidth limitations in wireless sector stymies the
flow of information to the customers.
The focus of the company’s effort is to develop a relational database that will help to search for
specific components of content available in a repository.
6.13 Broadstream (http://www.broadstream.com/)
BroadStream has developed software tools that measure usage patterns and quality of service as
experienced at the consumer end. Currently, content providers have no other means, besides the
log files of Content Delivery Networks, to determine both the usage and the quality of
transmission. Inevitably, the data available from log files is unreliable and does not provide
adequate information.
The rate of adoption of streaming media is low because consumer experience is not yet
satisfactory. Bandwidth limitations are not the only cause of poor quality of experience with
streaming media; standards based MPEG players permit narrowband transmission as well.
Content Delivery Networks can improve the quality of transmission if service level agreements
were monitored.
Furthermore, content providers need data on usage and the quality of reception to estimate
consumer willingness to pay for programming and to enforce copyright laws. They can implement
pay-per-view programs if they knew the data consumed by individual customers and price
according to the quality of reception.
Streaming media provides recourse to enterprises to reduce their costs of travel incurred for
training and corporate communications. Some companies like Hewlett Packard and Cisco have
taken to streaming media in the current economic slowdown. Many other enterprises are unwilling
to follow because they do not want to risk disruption in their mission critical applications. In the
absence of real time data on bandwidth usage, bandwidth utilization rates are unpredictable
which can jeopardize mission critical applications. Several companies, therefore, prefer to shut
down less important applications.
The technologies for load balancing that ensure the consistency in the quality of transmission of
HTTP traffic were not designed to achieve the same purpose with streaming media. A page once
viewed remains in the browser and its quality is unaffected by subsequent congestion on the
Internet. If another page is requested, data can be routed from the least congested server.
Consumers who connect to a media server, on the other hand, experience deterioration in quality
as the numbers watching a program increase. BroadStream’s Media sensor measures the quality
of reception in real time and BroadStream’s Fusion Center redirects traffic to a server best able to
serve the demand after it receives the information.
Currently, log files available from Content Delivery Networks aggregate information on data
served from several servers spread over a location. The reports generated from such data are
available after a lag of time. It is difficult to interpret the data because it does not relate to any
specific customer. Finally, the log file data matches the data transmitted to data received rather
than the data received to the bandwidth.
BroadStream’s Media Sensor helps to gather information about each customer, who remains
anonymous and is identified by a number, and gathers information about the bit rate, quality of
service and extent of usage. The data is gathered in real time, which aids capacity planning on
the fly.
Currently, content delivery networks decide on storage of content on the basis of probability of
usage on the edge sites. Consequently, content is often replicated in more edge sites than is
necessary. The data on usage, collected in real time, can be used to route traffic to the edge
when it is required so that a great deal of replication can be saved. However, software to take
advantage of the data gathered at customer sites is not yet developed.
The information on quality of service can be used to route traffic from a broader selection of
content delivery networks than is the case now. Currently, customers prefer reputed
organizations like Akamai, Digital Island or I-Beam. Content providers can afford to be agnostic
about their vendors if had a way to measure the quality of their service. They would then be able
to manage traffic optimally and reduce their costs.
6.14 Vicinity (www.vicinity.com)
Vicinity, a start-up offering enterprise location based services, is focused on providing licensing
and hosted services to multi-location companies such as retail chains, hotels and automobile
companies. Customers use the licensed software to provide location information, on their web
sites, to their clients. Two of its data centers also offer similar services off-site from its own web
site (http://www.mapblast.com/). License and hosting revenue accounted for $16 million of the
total of $ 20 million that the company generated in fiscal year 2001.
Although corporations have been purchasing Vicinity’s services, the beneficiaries are their
customers rather than employees using them for operational purposes. The staple of its
services are web based business finder, maps, driving directions and directory services, which
accounted for 76% of the revenue in the nine-month period ending April 30, 2001, 65% in fiscal
2000, 86% in fiscal 1999 and 91% of our revenues in fiscal 1998. The declining share of these
services reflects a trend towards an increasing emphasis on enterprise services.
Currently, a typical customer of Vicinity is Federal Express, which provides information on drop-
off locations on its web site for customers. The acquisition of over 300 of such multi-location
clients has prepared the way for developing enterprise services. In the immediate future,
Vicinity will offer data on supplies available at each of the centers of its customers such as
GAP. The data volumes are expected to rise from tens of Gigabytes to Terabytes with the
addition of these services alone.
Spatial intelligence is another enterprise opportunity that is available to Vicinity as a result of
recent customer acquisition. A typical application is to analyze the location aspects of matching
supply and customer need. Wells Fargo bank, for example, is interested in finding out the
distances between its branches and Coca Cola’s outlets in individual cities and neighborhoods.
Meanwhile, the established vendors of CRM applications like Siebel have extended their
capabilities into wireless space. This presents an opportunity for Vicinity because it has skills in
geo-coding that CRM companies need before they can offer location-based services. Geo-
coding under girds services such as time proximity search or estimate of time required to move
from one location to another.
Over the longer run, Vicinity is betting on location based field service automation services.
However, mission critical services of this nature require deployment from behind the firewall
acquisition of additional technological capabilities to integrate with related enterprise
applications.
Currently, its development effort is focused on usability aspects of its products. Although,
Vicinity is not looking to invest in integrating imagery information, it is working on the quality of
its vector data. It purchased BeeLine in 2000 to improve the rendering of maps. Customers will
now be able to zoom in and zoom out using the new capabilities.
Positional accuracy and location awareness is another aspect of improving the usability of
location-based services. Currently, customers feed their zip code information, which covers an
area much too large for customers who use the services to save time, resources and effort.
Vicinity has partnered with Signal Soft to add sensors to cellular phones, which will identify the
location within a 100-meter range. Customers will be presented with options, on any aspect of
their choice, available within that location.
However, positional inaccuracies will continue to be a thorn in the development of enterprise
applications. Vicinity’s geo-coding is able to estimate the spatial parameters of a destination
with an error rate that does not exceed 7 square foot. Location identification has ways to go
before it can match that accuracy.
Although Vicinity has incurred huge losses of $27.3 million (including a $ 7.6 million one time
charge) in fiscal year 2001 compared to $ 16.5 million in 2000, it has cash reserves of over $
94 million to continue the development effort. It will also benefit from the completion of the
development of an Enterprise Location Server that helps to increase the reliability of the
services offered.
6.15 Mediabin (www.MediaBin.com)
Mediabin is positioned to meet the brand asset management workflow needs of large enterprises
such as Ford Motor Company. Its software automates the tasks of filing, indexing, retrieving and
reformatting images and related marketing assets for a variety of media such as print and the
web. Assets are centralized in a repository that serves as a clearinghouse for users, inside the
enterprise for use by all databases, and by partners. A key component is the metadata that helps
users, whatever their preferred tool, access the assets.
Digital asset management (‘DAM’) software for rich media management has been available with
entertainment and printing/publishing companies but has not been deployed beyond these
vertical markets. MediaBin sees a more compelling need for managing the flow of brand-related
content within a large multinational company.
MediaBin’s marketing strategy is intended to save customers the cost and the complexities that
they experience in preparing content, especially images, and in deploying brand assets within
and beyond the enterprise. Currently, customers have to pay in excess of $100,000 to purchase a
license for a content management tool. MediaBin chooses to provide an infrastructure that is
focused on the management and deployment of digital brand assets. The company’s image
management and its support for standards (especially XML, and COM, Java and PERL APIs)
allows easy interface with other systems. The license for MediaBin’s software begins at $50,000
and the price increases with complexity. MediaBin’s portal type of software, with a directory of
files and file management tools, can be deployed within a day as long as a file structure is already
in place.
Large enterprises like the automobile companies have to cope with a welter of brand related
collateral as they offer numerous models in the market place. Images and other related
information lies scattered with advertising firms, marketing agencies and market communications
departments within the enterprise. Within a company, several departments create content which
they need to share. A repository reduces the workflow between these departments. MediaBin has
developed a platform to create the applications to manage images centralized in a repository.
The need to reuse images has grown as the number of channels of communications and
participants has grown, due mostly to the emergence of the web. Images have to be rendered in
a variety of formats such as JPEG, TIFF, PDF, etc., besides innumerable shapes, colors,
resolutions and sizes. In more recent times, MediaBin has had to respond to the demand for
version management of images.
The latest upgrade version of MediaBin’s software graduates from workflow management for a
workgroup to enterprise wide management of brand assets. It has enhanced its software to scale
and has broadened the range of file types that can be previewed via a browser, to include PDFs
and MS Office documents such as PowerPoint presentations. In order to handle larger volumes
of traffic, the software enables management from a cluster of servers each of them specializing in
specific types of traffic, e.g., data flow from images comes in larger blocks than related textual
material.
Metadata, and its incorporation with the media file in an XML “wrapper,” plays a key role in the
enterprise version of the software. It serves as the intersection point for content to flow between
departments and users to search for the content they need.
Security features had to be enhanced to enable access for especially participants outside the
enterprise. The browser has been redeveloped to allow access and downloading of assets after
password checks.
MediaBin has been encouraged to integrate media related workflow processes in the entire
enterprise by changing attitudes of the IT departments. The large volumes of rich media-
generated traffic were anathema to IT departments. Lately, the increasing numbers of customers
of images within the enterprise has forced a reappraisal with IT departments.
6.16 Scale Eight (www.scale8.com)
Scale Eight is a start-up offering Internet based storage services with a niche market in the rich
media segment. Unlike the incumbents, such as StorageNetworks and Storability, Scale Eight
serves clients who prefer to outsource the management of all storage services outside the
firewall.
The hallmark of Scale Eight services is that file management happens locally on a web browser
or on a LAN within an enterprise. Customers also have the option to mirror their file systems in
several locations spread around the world with its Global Storage Port device. Scale Eight’s
services are, therefore, particularly useful for collaboration within enterprises with subsidiaries
around the globe.
The Global Storage Port device is attached to the local area network and the file system is
cached locally at customer premises. Customers manage storage devices resident on the four
central locations, in San Francisco, Tokyo, London and Virginia on Scale Eight’s network.
Scale Eight does not cater to the security sensitive applications that have to be offered on the
premises of enterprises. By creating an infrastructure outside the enterprise on the Internet, Scale
Eight does not have to work with proprietary software and associated equipment available from
costly vendors like EMC. Scale Eight uses commodity equipment, JBODs, and standard software
to drive down the costs to a fifth of the incumbents. The current price of its service is $30 per
managed mirrored gigabyte compared to $150 from competitors.
Predictably, Scale Eight’s sweet spot is personal storage, including executive baggage, which
includes mostly music files from MSN music services besides digital photographs, digital images
and video. Overall, the market is split into half each between the consumer and the business
segments. The market for visual files is divided equally between digital photography/images and
video. The company is also exploring wireless markets in European markets but does not see an
immediate prospect in the USA.
7 CONCLUSIONS: IMPLICATIONS FOR BUSINESS STRATEGY
Storage area networks are analogous to radar systems in the airline industry whilst wireless
technologies are like individual aircraft. Just like data from individual aircraft are received at a
central point to decide on landing and take-off schedules for aircraft, storage area networks
accumulate information and churn out alerts for the mobile staff of an enterprise. Similarly,
wireless devices like airplanes receive data from a central point about possible dangers to them
en route. Enterprises would find as hard to function as airlines would without their radar
systems; they will cease to be agile enough to be relevant in the marketplace.
Storage industry’s transition from directly attached storage and LAN-attached storage to
storage area networks has far reaching ramifications for information management and business
strategy in the near-term future. The value of information is dramatically enhanced when large
volumes of related data are gathered and correlated at a single point to draw out significant
patterns for decision making purposes. Its utility is also augmented by ease of search of
information at a granular level so that information can be reused. Finally, the value of
information rises as it is shared more widely in a variety of situations. Storage technologies,
especially network storage, have the ability to achieve all these goals.
The boundaries separating sources of information, such as their medium of publishing, location,
computing systems, will begin to wither and all data will be flow into colossal repositories.
Businesses will be able to use not only textual information but also imagery, moving pictures,
photography and voice data to their competitive advantage. Imagine the impact of catalogs with
not only still photographs but also movie footage to demonstrate the capabilities of a product
like digital cameras to on-line buyers who may not have previous exposure to them.
Multi-media communication is creating new possibilities for managing work processes. The
engineering industry, for example, can use a combination of imagery, design graphics and
textual data that has to be often exchanged between field staff engaged in construction and the
planning staff working from offices. The entire documentation, accessed from a single source,
can be shared by all staff members working on-site.
LAN attached storage devices were a help at a time when enterprises needed to raise the
efficiency of work groups. Storage area networks allow enterprises across several locations to
collaborate with each other. A striking illustration is the advertising industry which often has
clients spread over several locations or its value chain includes shooting, production and
distribution spread over several locations worldwide. The availability of work-in-progress on
storage area networks helps the groups working in different locations to collaborate since time
lags in receiving the material is reduced to virtually zero.
Storage technologies will play a critical role in the commercial viability of a whole new range of
products and services by providing capacious capacity for storage. Services such as digital
photography and video-on-demand will simply not be possible without large volumes of
storage. Unified Messaging has been unaffordable for the large majority of potential customers
because of its high cost. Storage technologies, aided by software tools for management of
capacity, can substantially lower the costs of managing messages.
Similarly, the commodious storage capacities of storage area networks have aided the rise of
the voice information industry. Speech recognition technologies, create large files that would be
hard to use without the cavernous storage capacity of SANs. The very same files can be
mirrored on several sites for convenient access locally. The prohibitive costs of maintaining call
centers can be lowered by automation enabled by speech recognition technologies besides
rescuing customers from the “menu-hell” of interactive voice recognition systems.
The distributed nature of storage area networks creates new channels that can diversify the
revenue sources. Revenue realization for movies is hamstrung by the inability to cater to
smaller size audience. A striking illustration is the grossly underserved education market for
documentaries. Movie halls are not amenable to viewing on demand. Similarly, car owners
require content on demand unlike the preferences of mass audiences. Storage area networks
have the ability to serve customers on demand and to view content in their preferred ways.
Channel diversity is not limited to the flexibility in distribution of content to audience types. The
users of content can be differentiated by the diversity of devices especially among owners of
wireless handsets. Typically, format needs differ such as the standards used for playing movies
and the language. Storage area networks, aided by related content management tools, can
convert content from one format to another or from one language to another in real time
allowing its reuse.
Corporate strategy, especially decision making, will be fundamentally transformed by the
capabilities that storage technologies afford. Storage technologies enable a quantum leap in
the volumes, velocity and variety of information so that the quality and speed of decision
making is greatly improved. The shorter lead times enable companies to make price
adjustments to supply and demand imbalances in near time. Errors in decision making will be
reduced as companies have the ability to process inter-related information, compare their
metrics with best practices and to check with their past experience.
Correlations between pieces of information are possible when the data is defined abstractly
before relationships can be drawn. Metadata, or the data describing data, plays a crucial role in
discerning the relationships in disparate pieces of very large volumes of information. A great of
health information is meaningful when it can be related with patient information. Similarly, all
location information can be interpreted when it is related to spatial co-ordinates. Past
information on a patient, record of illnesses in youth, is useful in making decisions about
diagnosis of a person when he or she grows old. None of this would be feasible if the inter-
related information cannot be retrieved and presented in an intelligible format to the user in real
time.
Agility will be the byword of enterprises who take advantage of storage technologies. Field
forces and sales staff especially will benefit from time saved on paperwork and the ability to
access information to service consumers’ faster. The maintenance of far flung assets will be
faster as field forces navigate to their specific destinations aided by graphically rich maps that
they can view on their handheld devices. They can review technical information on the tip of
their fingers to service their clients. Sales personnel will not be burdened by the document
luggage; they can print remotely accessed documents on customer sites. They can also be
alerted to opportunities, e.g. trucks can be alerted about parcels they can pick up by delaying
their departures.
The agility of enterprises will greatly increase as business processes are streamlined by
reducing paper work and duplication in document generation. Documents are often referred by
several participants in an enterprise; the consequent paper shuffle causes duplication at
several levels and slows down decision making as information flows from one point to another.
The health industry is a classic example of an industry that is bogged down by paperwork.
Once documents are available centrally on storage area networks, they can be viewed
conveniently much like any information is on the Internet.
Enterprises create brand related imaging and related collateral which can be shared rather than
duplicated when it is stored on a device. Advertising agencies can, for example, keep their
work-in-progress on a storage area network so that their clients can plan their campaigns.
When the collateral is ready, it can be published not only for printed publications but also on the
web. A great deal of duplication takes place because the collateral has to be published in a
variety of shapes and sizes, different resolutions and color shades. These conversions can
happen quickly and with low incremental costs if the template is stored at a single point.
Network storage is evolving into a grid which would be as ubiquitous as electrical networks in
the near term future. Storage providers are integrating fragmented networks by joining them
with the abundant dark fiber, converted into optical networks, widely available in the USA and
elsewhere in the world. Pioneering companies like Giant Loop and In Range Technologies
have set the stage for global storage area networks.
The process is completed by logical representation of the physical capacity by virtualization
software before centralized management of the entire network becomes possible. A bevy of
start-ups such as Datacore Software, Legato and Falconstore supported by financially strong
companies like IBM will accelerate the process of centralized management of storage area
networks.
The full benefit of storage area networks will be realized when software for storage
management is deployed with success. Storage hardware, installed in the initial stages, has
been underutilized because software for its optimal utilization was not widely adopted.
Economic pressures of the recent recession are driving companies to make greater use of
software to manage their storage hardware as a single network and to distribute the load
evenly over all disks and tapes.
Initial progress in tying storage devices has been limited to homogenous devices within the
enterprises for security reasons. Over time, heterogeneous devices will also be inter-connected
in a single storage area network as companies agree to share APIs for inter-operability. For
security reasons, networks will not be entirely open but some segments of the industry catering
to needs such as digital photography and imaging will take advantage of bargains available
from outsourcing solutions.
The speed of retrieval of information is limited by the inability to increase the speed of operation
of disks. This will be mitigated by parallel processing that is possible by dividing information
flows into several streams of data. File systems are increasingly able to manage parallel
streams of traffic.
Widespread adoption of SANs depends a great deal on the quality of service available to
customers. In a shared environment, companies can experience inconsistency of performance
as clients make conflicting demands on storage capacity. This would be unacceptable to I/O
intensive applications like data warehouse applications that have to offer consistent
performance such as query time.
Cost reduction will be another major driver of greater adoption of storage area networks. Disk
drives are way too costly for many of the applications especially where images are utilized.
Tapes present a viable alternative to disks to reduce costs especially if their latencies can be
lowered. Near-line tape storage promises to increase the adoption of tapes to substantially
reduce the costs of managing imaging data.
Storage technologies not only enable new applications their commercial acceptance is also
dependent on the progress of applications. In the near term future, several applications will
come to fruition after years of nurturing. Digital photography is one of the major new
applications that create storage needs and an application that lends itself to mass usage. The
conditions for its widespread adoption are available as traditional photographs can now be
digitized, digital photographs can be processed conveniently at kiosks, work with common
software such as Windows XP and communicated over the Internet. In addition, agreement on
metadata standards will enable reuse of photographs. Storage demand will increase in step to
enable sharing of digital photographs.
Similarly, medical applications of storage have long been dormant because clinical information
systems were resistant to automation. The primary reason for slow progress in the health
industry was the lack of consensus over the vocabulary of medical terms, its logical
representation and the metadata. These barriers have been overcome so that data integration
in medical enterprises is now possible. Health enterprises can substantially reduce opportunity
costs since duplication in paperwork is common and time delays are extremely costly when
expensive equipment is idle. In addition, the speed and quality of decision making will greatly
improve as the volumes, velocity and variety of information available increases aided by
storage technologies.
Location based services will grow as the initial spade work of rendering map data has been
largely completed. Increasingly, LBS companies will correlate map information with related
attribute information besides imagery and aerial photography to offer premium services.
Storage demand will increase especially as media intensive services are offered. Enterprises
are likely to adopt the more advanced services available.
Vehicle Telematics is growing out of its infancy as entertainment and information services are
offered besides the staple safety and security services. Voice information services will greatly
facilitate the delivery of information services. Mobile professionals are likely to be receptive to
information services that they require in the daily routine of their business. Satellite audio
entertainment has already been launched and its success will encourage several other
companies to offer entertainment services to vehicles.
Business Intelligence and Customer Relationship Management data warehouses have an
established market which will grow in size as the variety of information gathered increases. In
addition, the adoption of BI/CRM will be a boost to the growth of related services such as
location based services and remote document retrieval.
Storage area networks will increasingly drive the adoption of wireless technologies as the
backlog of integration of work processes and information technologies utilized by storage
technologies is completed. Wireless technologies will improve communication with customers
and the quality of services available to them. For a low incremental cost of integrating wireless
technologies, mobile computing will reap the gains from the enormous investments in storage
technologies.