John Waclawsky Ph. D. Services Architecture and Governance Motorola, Inc. From: Heavily Centralized Control Paradigms To: An Increasing Decentralized World via Internet and Web Technology Titan against Titan: Titan against Titan: What Technology will What Technology will Win? Win?
29
Embed
John Waclawsky Ph. D. Services Architecture and Governance Motorola, Inc. From: Heavily Centralized Control Paradigms To: An Increasing Decentralized World.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
John Waclawsky Ph. D. Services Architecture and GovernanceMotorola, Inc.
From: Heavily Centralized Control Paradigms To: An Increasing Decentralized World via Internet
and Web Technology
Titan against Titan: Titan against Titan: What Technology will What Technology will
Common mechanisms are good …for applications too? 1. Some applications can leverage standards …billing etc.
2. Belief: Common control into the application space will facilitate interoperability, easier application creation, more application utility and numerous new applications will emerge by extending commonality. This is a common perspective of IMS/SIP advocates BUT: has IMS/SIP led to any new applications?
“differentiation IS the game”.... Geoffrey Moore
Competition and Commonality Standards vs. De facto
Applications drive technology usage, not the selection of some common protocol or standard.
Competition and Commonality (continued) Standards vs. De facto
Smell Test: Will competition stop? …a single solution /application / signaling / control / format / data protocol, or any other common way to serve customers in a non-competitive manner…
de facto: Un-commonality is standard for applications
•Standards typically commoditize products tend to make products and services look more or less alike
•Standards may be giving competitors some control or even veto power•Applications don't want to “talk” to each other for business reasons
•Innovators always look beyond standards for ways to lead
Model extended because:•Accelerating technology changes •Disruptions and redefinition •Relentless on-going innovation•Business decisions are colored by:
•Politics/Ideology, •Financial considerations •Technology religion (driven aspects of a company’s or even an individual’s personality).
L8 - Revenue and Profit
L9 - Politics
L10 - Technology Religion
The upper three layers are mainly about competitive issues
Everyone needs the bottom four layers of the OSI modelThe split is over how to exploit the top of the extended OSI model
Incentive to follow successful lower layer standards and, as a result, allow network-connected products and services to enjoy access to the widest audience
Create new standards to extend connectivity when new technologies emerge or provide ways to better leverage the internet, such as WiMAX
“connectivity is its own reward” was often echoed by the early Internet participants, and is embodied in Metcalfe’s law
GOALS: • Standardize communication, NOT application behavior or control of end users. • Everyone to benefits from connectivity.
RESULTS: • Experimentation for new applications, services and technology exploded •Innovation breeding ground spawning numerous high-market capitalization companies: Amazon, Google, eBay…•Enormous wealth engine - February 6th 2006 SIP Forum[1] presentation that concluded “The Internet is responsible for the largest creation of shareholder value in the shortest time in history.” [1] http://www.sipforum.com/
GOALS: •Standardizing communications including: •Interoperability between applications in their respective vertical markets, • End-user control • Total control of application behavior.
RESULTS: Meeting goals rooted in existing thinking about networking •A highly-controlled, but much-reduced experimentation environment •Depressed innovation activity•From our innovation migration lessons, it is becoming more apparent the trend is that the Internet is taking over
Area of Common Benefit The standardized lower levels have also helped solve the bootstrap problem for innovators. These layers facilitate the spread of new,
unconventional products and services at the higher layers of the protocol stack. Via existing standardized lower networking layers, anyone can
now download and install the software needed to use such new innovations driving concepts such as social networking. That's a key
reason new innovations can reach critical mass so quickly.
System technologies about control: • IMS• Quality of Service (QoS)• Deep Packet Inspection (DPI) • RST Injection for TCP protocol• Session Initiation Protocol (SIP)• Digital Rights Management (DRM)
Control technologies are mainly desired by companies in the restrictive competition eco-system but have little value for the end-user customers.
Consider an evolution about relationships
• Mash-ups• P2P • Encryption • People technology
• Creating: Blogs, user generated content, podcasts• Connecting: Social networks, virtual worlds• Collaborating: Wikis and Open Source• Reacting to others: Forums, Ratings, Reviews• Organizing content: Tags• Staying aware: RSS, widgets and Twitter
Many of these technologies have demonstrated considerable end-user value (for example, Bit Torrent, Skype, etc.) but many provide little or no value to the restrictive competition eco-system.
Other related issues:1. Infrastructure costs!2. Privacy concerns! 3. Missing services/functions?
Another Example: Circuit Voice vs. VoIP What about Lawful Interception (LI)?
Centralized: Circuit Switched network is easy Data network: Session Border Controller (SBC)* as the point of convergence for VoIP packets.
Implementing LI on SBC is the VoIP equivalent of wire tapping on a circuit switched network.
*SBC is typically a VoIP session aware device that governs the manner in which VoIP calls are initiated, conducted and terminated in a network.
Distributed: VoIP IP provides numerous methods to ensure data security. no standardized manner to distinguish voice packets no telling which path the IP packet will take what headers get added.
Decentralization is effecting LI too!
BTW: this is BTW: this is all all
true for any true for any kind of kind of traffictraffic
System technologies about control: • IMS• Quality of Service (QoS)• Deep Packet Inspection (DPI) • RST Injection for TCP protocol• Session Initiation Protocol (SIP)• Digital Rights Management (DRM)
Control technologies are mainly desired by companies in the restrictive competition eco-system but have little value for the end-user customers.
Consider an evolution about relationships
• Mash-ups• P2P • Encryption • People technology
• Creating: Blogs, user generated content, podcasts• Connecting: Social networks, virtual worlds• Collaborating: Wikis and Open Source• Reacting to others: Forums, Ratings, Reviews• Organizing content: Tags• Staying aware: RSS, widgets and Twitter
Many of these technologies have demonstrated considerable end-user value (for example, Bit Torrent, Skype, etc.) but many provide little or no value to the restrictive competition eco-system.
Other related issues:1. Infrastructure costs2. Missing services/functions3. Privacy concerns
System technologies about control: • IMS• Quality of Service (QoS)• Deep Packet Inspection (DPI) • RST Injection for TCP protocol• Session Initiation Protocol (SIP)• Digital Rights Management (DRM)
Control technologies are mainly desired by companies in the restrictive competition eco-system but have little value for the end-user customers.
Consider an evolution about relationships
• Mash-ups• P2P • Encryption • People technology
• Creating: Blogs, user generated content, podcasts• Connecting: Social networks, virtual worlds• Collaborating: Wikis and Open Source• Reacting to others: Forums, Ratings, Reviews• Organizing content: Tags• Staying aware: RSS, widgets and Twitter
Many of these technologies have demonstrated considerable end-user value (for example, Bit Torrent, Skype, etc.) but many provide little or no value to the restrictive competition eco-system.
Other related issues:1. Infrastructure costs2. Missing services/functions3. Privacy concerns
• Network coding is a field of information theory and coding theory and is a method of attaining maximum information flow in a network
• The core notion of network coding is to allow and encourage mixing of data at intermediate network nodes.
• In contrast to traditional ways to operate a network that try to avoid collisions of data streams as much as possible • A receiver sees these data packets and deduces from them the messages that were originally intended for the data sink.
• This is an elegant principle that implies a plethora of surprising results
System technologies about control: • IMS• Quality of Service (QoS)• Deep Packet Inspection (DPI) • RST Injection for TCP protocol• Session Initiation Protocol (SIP)• Digital Rights Management (DRM)
Control technologies are mainly desired by companies in the restrictive competition eco-system but have little value for the end-user customers.
Consider an evolution about relationships
• Mash-ups• P2P • Encryption • People technology
• Creating: Blogs, user generated content, podcasts• Connecting: Social networks, virtual worlds• Collaborating: Wikis and Open Source• Reacting to others: Forums, Ratings, Reviews• Organizing content: Tags• Staying aware: RSS, widgets and Twitter
Many of these technologies have demonstrated considerable end-user value (for example, Bit Torrent, Skype, etc.) but many provide little or no value to the restrictive competition eco-system.
Other related issues:1. Infrastructure costs2. Missing services/functions3. Privacy concerns
QoS How can QoS work today and in the future? ….when you consider…
• Emerging future: overlay techniques (P2P), mash-ups, traffic scattering, network coding. • Encryption or use packet-obfuscation
Lowest prioritization for all encrypted traffic? – Privacy is systematically discriminated against.
• Most of the time the SERVERS ARE SLOW and NOT the network. • Low Utilization is a fundamental part of network design
Redundancy for reliability. Capacity for peak loads. What does it mean to run a link/box at 10%?
• Race with Moore's Law Link queue can empty faster than you can run instructions to make QoS decisions.
• QoS adds complexity Fiber capacity shifts bottlenecks from pipes to nodes and because of the enormous fiber speeds available, adding node queues to the mix of things that need to be QoS configured and managed doesn't appear to simplify the QoS challenges.
• Where is the ROI? • etc.
QoS is NOT an adequate substitute for capacity and potentially makes a bad
• It’s an environment that fosters experimentationClearly "the place" for innovation of communication services Seems to be about the absence of impediments
The lack of impediments seen in one eco-system and not the other appears to be making a huge difference in where innovation (and the associated wealth it generates) will be most successful.
• More experimentation then more luck! More $$$!
A major part of innovation is what we can call unexpected usage (or luck). However, the luck seems to be on the Internet side these days.
• Application-independent, TCP/IP or UDP are the backbones of the end-to-end nature of the Internet.
If history is any guide, a betting man would probably look for the next large market cap company to be about services and
A Major Challenge for the Restricted eco-system Technology….
How can any technology which relies on extensive core network control and takes an application focus and consider packet information invariant, adapt to overlay techniques found in P2P networks, traffic scattering, network coding, the increasing use of encryption, the emergence of cloud computing, as well as trends related to dynamically composed and instantiated concoctions (formally known as applications) at the edge of the network?
The web is becoming “THE” programming development platform. Now, many view the web as the ultimate programming platform
Early, half-baked is rewarded better striving for perfect is the enemy of good, and doing so is very
time consuming, very expensive, and easily by-passed Everyone wants to differentiate their products People always dream of reaching de facto nirvana Lock in your customers
mine your customer set with derivative products and advertising;
Politics (or group affiliation) overrides many choices Economic incentives to succeed in the market are the
major goals tied to differentiation strategies Technology religion (personality preferences) will
override the benefits of standards to product developers and people running companies focused on success.