Michael L. Gonzales HandsOn-BI, LLC Quarter 1, 2005download.101com.com/pub/tdwi/files/HandsOn-TechUpdate.pdf · Cognos As leading RDBMS vendors blend relational and multidimensional

TechUpdate is published quarterly and is available exclusively at www.tdwi.org.

By: Michael L. GonzalesHandsOn-BI, LLC Quarter 1, 2005

See Technology Update Live! with Michael L. Gonzales atTDWI’s Spring and Fall World Conferences.

For more information about TDWI events visit our Web siteat www.tdwi.org.

2

Table of Contents

Introduction............................................................................................................... 3HeadsUp .................................................................................................................. 5

Overview ............................................................................................................................ 5HeadsUp ............................................................................................................................. 6

Database Management Systems ............................................................................. 7Data Integration........................................................................................................ 9

Data Quality...................................................................................................................... 12Data Profiling ................................................................................................................... 12

OLAP...................................................................................................................... 13Query and Reporting.............................................................................................. 16Dashboards ............................................................................................................ 17Data Mining ............................................................................................................ 19Metadata ................................................................................................................ 21Appendix A –Glossary........................................................................................... 22Appendix B –References....................................................................................... 24

Tables

Table 1 –Big Three Overview.................................................................................. 5Table 2 - HeadsUp ................................................................................................... 6Table 3 –Database Management Systems ............................................................. 8Table 4 –ETL......................................................................................................... 10Table 5 - OLAP....................................................................................................... 14Table 6 - Query & Reporting .................................................................................. 16Table 7 - Dashboards............................................................................................. 17Table 8 - Data Mining............................................................................................. 20Table 9 - Metadata ................................................................................................. 21Table 10 - Glossary ................................................................................................ 22

3

Introduction

As an active BI practitioner, I’ve often searched for perspective and insight on the latest and greatest technologies and techniques in our industry.Unfortunately, to find any decent information often requires you to purchasethe report. And whether I pay $90 for a single page or $3000 for a 20 pagereport, I’ve often found limited value for the money spent. Therefore, thisquarterly update is free. That doesn’t mean contributors and I did not go through great effort to gather, synthesize, and document the contentpresented, because we did. But I want this publication to complement theresearch activities of peers and fellow practitioners—and to do so without anyrequirement of membership or costs.

The report is divided into the core sections outlined below:

Database Management Systems Data Integration OLAP Query and Reporting Dashboards Data Mining Metadata

A panel of seven individuals contribute to the content of the paper including,three representatives from leading BI vendors, three senior IT representativesfrom companies with large BI initiatives, and me. Each of us contributesperspectives for all the components identified above. These perspectivesaddress two time horizons, a short-term and a long-term. The short-term isone to two years from today, emphasizing the current trends and best-in-classtechnologies. And the long-term view examines components more than twoyears into the future. As the author, I not only contribute my own perception,but gather and blend all perspectives into a single view for each component,and formalize that view in this report.

Since this is a high-level, summary view of the components defined above, notall products are covered. However, products that are not mentioned, does notmean they are irrelevant technology. And those that are mentioned in thispaper should be regarded only as representative of the components.

Finally, this publication is designed with the idea that most of us do not havetime to read through volumes of material for any insight. Instead, we want the

4

insight made readily available, pushed right in front of us. To that end, themost salient information offered is found in the tables within each componentsection. If you want to get to the heart of the information provided, it will bethere in brief, terse commentary, best used to stimulate further questions andresearch. It is my sincere hope that you find a nugget or two of value fromthe report.

5

HeadsUp

This section is designed as an abstract to give readers a quick, summary viewof the report content. Here you will find an overview of the big threecomponents covered in the report: Database Management, Data Integration(ETL), and OLAP. Moreover, a HeadsUp subsection is included that draws thereader’s attention to products that warrant further investigation due to the technological prowess exhibited in their most current releases.

Overview

Of the seven technical components covered in this paper, this section providesa high-level view of data management, data integration and OLAP. We refer tothese components as the big three simply due to their share of BIenvironments.

Table 1 - Big Three Overview

Key Players Outlook

DatabaseManagement

IBMMicrosoftOracle

IBM, Oracle and Microsoft will deliver BI platformswith increased capabilities including ELT, OLAP,Mining, and Reporting. They will focus on TCO.

DataIntegration

AbinitioAscentialIBMInformaticaMicrosoftOracleSAS

Pure play ETL vendors will face stiff price pressureand competition from RDBMS leaders such as Oracleand Microsoft. Downward pricing pressure fromRDBMS vendors will result in greater acceptance ofCOTS packages, but no new player will enter thespace.

OLAP BusinessObjectsCognosHyperionMicrosoftMicroStrategy

TCO and downward pricing pressure from RDBMSleaders will result in a growing market share forthese vendors at the expense of traditional, pureplay OLAP vendors. Pure play OLAP will be forcedback to special applications such as budget andforecasting. MDX is the accepted interface, whileXML/A flounders.

6

HeadsUp

HeadsUp identifies and acknowledges products that demonstratecharacteristics of leading technologies with real, tangible application in the BIspace. Readers should track these products and/or investigate their potential.They are chosen purely from my perspective, based on my experience in theindustry and exposure to a wide variety of technologies.

Table 2 - HeadsUpProduct CompanyMapIntelligence

Integeowww.integeo.com

End-users can finally realize the practicalapplication of blending spatial data andanalysis with traditional BI technology. Thisproduct minimizes the technical knowledgerequired to build a spatial-BI application.

PolyVista PolyVistawww.polyvista.com

This product exudes the phrase--leadingtechnology. When it comes to seamlessintegration of data mining, OLAP and 3-Dvisualization, PolyVista stands alone.

SAS ETL Studio SASwww.sas.com

SAS offers data mining and statisticalanalysis technology with a solid reputationfor performance. When this technologyserves as the foundation for an ETL tool,bona fide data quality, data cleansing, andinformation enhancement are achieved.

OracleWarehouseBuilder

Oraclewww.oracle.com

All leading RDBMS vendors have enteredthe ETL space. But none have been able toprovide a viable ETL offering until Oracle’s Warehouse Builder version 9.2.

DMExpress Syncsortwww.syncsort.com

An effective ETL technology built on a battletested performance engine, DMExpress is aviable ETL alternative, scalable from alaptop to a 64-way Superdome.

HyperionDashboardBuilder

Hyperionwww.hyperion.com

Technology that epitomizes the notion ofcomplex applications made simple toimplement. The product virtually eliminatesthe need for programming in order to buildand deploy complex dashboards.

7

Database Management Systems

The current database market is dominated by three players, Oracle, IBM, andMicrosoft. All three are trying to extend their market position by adding suchfunctions as BI integration (OLAP, Data Mining), complex data type andmessaging support (XML, JMS and EJB capabilities), and dataquality/integration (ETL).

Within the next few product release cycles, we will witness each of the threemajor vendors greatly strengthen their products in several ways, including:

ETL -- The in-database ETL technology will continue to mature, becoming solidtools for traditional ETL functions.

OLAP -- Each of the three database vendors will continue maturing their ownmethod for servicing the relational, as well as, multi-dimensional needs of theircustomers.

Data Mining -- With the advent of Predictive Model Markup Language (PMML),building may continue outside the enterprise database of choice, but theexecution mining models will be done in the database.

Messaging & Active Data Warehouse -- All leading RDMBS vendors offer theability to provide messaging, a building block for web services andfundamental to establishing a Service Oriented Architecture (SOA). Moreover,each of these vendors currently makes available the technology forimplementing the concept of Active Data Warehousing.

Although these leading database vendors may not achieve best-of-class for thespecific technology categories above, they will offer sufficient functionality.And, when you combine that functionality with their price/bundling leverageand the ability to control and administer all within a single enterprisearchitecture, the overall value is significant.

It is entirely probable that over time, the database may be the soleinfrastructure component in the new product class of data infrastructure(storage and integration together). Refer to Table 3.0 for more information.

8

Table 3 - Database Management Systems

KeyPlayers

1-2 Year Outlook +2 Years Outlook

RDBMS OracleIBMMS SQL

RDBMS vendors will continueintegrating BI functionality.Support for self-management, mixedworkloads for tactical andstrategic requirements, anda focus on TOC.

Leading RDBMS vendors willintegrate ETL, OLAP, Mining,and Reporting in a BIplatform for enterpriseinitiatives. Increasedattention to GRIDarchitectures and continuedfocus on TOC.

MDD MicrosoftHyperionCognos

As leading RDBMS vendorsblend relational andmultidimensional data into asingle environment, the needfor proprietary MDD vendorswill continue to decline inpopularity except in specificapplications such asplanning, budgeting, andforecasting.

Very few pure OLAP vendorswill remain. Real-timeanalytics will drive newdefinitions of OLAP resultingin new architectures thatmange streaming data.

OpenSource

MySQL Open source RDBMSofferings will continue tomature with increasedpopularity, functionality, andcorporate acceptance.

Open source databases willcontinue to put pressure atthe low end, primarily inspecialty applications suchas Web servers.

Fast QueryorAppliance

SybaseIQNetezza

Will remain a nichetechnology.

Technology such as 64bitCPUs and concepts such asGRID computing, coupledwith plummeting hardwarecosts will all minimize anymomentum or significantgrowth.

9

Data Integration

Current best-in-class trends in Data Integration tools are representative of theevolved meaning of traditional Extraction, Transformation, and Loading. Manyof the leading vendors competing in the ETL space are aggressively expandingtheir features to address concepts such as near real-time data feeds and thesupport of a Service Oriented Architecture (SOA).

The objective of an SOA is exposing any capability (service) available withinthe data integration toolset to any application (consumer) that wishes to usethe functionality. This functionality is typically exposed as a web service orother messaging call. Although this concept promotes a very high degree ofreusability, it is only mature enough for low volume transactions at the presenttime.

This data integration trend places more focus on message technology andindustry standards such as JMS and EJB, essentially absorbing the role ofEnterprise Application Integration (EAI). Vendors such as Ascential andInformatica are leading the charge with this expansion of technology.

Also, the current trend sees leading vendors such as Abinitio, Informatica, andAscential establishing high performance transformation engines. Abinitio andAscential are also focused on dynamic parallel technology used for SMP, MPP,as well as, the beginnings of GRID computing. GRID is promising for low cost,high volume transformation. There are many who feel that GRID will be theenabler, allowing the enterprise to achieve a common data transformationlayer that will power all data integration and quality initiatives.

In the near future SOA is seen expanding into high volume data integrationtasks. This will allow extensive reuse of common components beyond the ETLframework. Also, inline capabilities such as data quality auditing, will beincluded within a single vendor’s product sets.

Looking further into the future, a single data integration framework fortransformation handling in the form of “unit of work”transactions will be thenorm. Refer to Table 4.

10

Table 4 - ETL

KeyPlayers

1-2 Year Outlook +2 Years Outlook

Overall Oracle, IBMMicrosoftInformaticaAscential

Oracle, IBM, and Microsoft will continue to mature theirETL offerings; while, Informatica and Ascential willcontinue to differentiate themselves by expanding into thearea of enterprise data integration.

Pure ETLProducts

AscentialAb InitioInformatica

HeadsUp:SAS

Growth in management ofsemi-structured data suchas XML, as well asincreased bundling offeatures for data qualityand profiling. But thisgrowth will be due toincreased acceptance ofpackaged ETL solutionsand downward pricepressure, not in newopportunities

No more than two to threemajor players will continueto fight a feature, function,pricing war in the face ofgrowing competition fromRDBMS vendors.

RDBMSETL

OracleMicrosoftIBM

HeadsUp:Oracle

Oracle will continue to leadand even gain marketshare as a result of itsacquisition of PeopleSoft,which will eventually switchits ETL product fromAscential to OWB.

Major improvements fromMS and IBM. The combinedofferings between theseRDBMS leaders will putprice and market pressureon pure play vendors.

RDBMS vendors offer strongtraditional ETL capabilities.Their increasingly capableofferings increase pressureon pure play vendors anddeter others from enteringor even maintainingpresence in this market.

11

(Table 4 continued.)

DataIntegrators

InformaticaAscentialIBM

EII technology will providefederation across strategic(DW, DM) and operationalsystems.

IBM will lead withfederation acrossstructured andunstructured data. Oracleand MS will take strongerpositions with federationtechnologies.

Considered important BIfunctions are the ability toanalyze unstructured dataand near real-timecapability.

What was ETL will now beconsidered data integrationservices. Technologies suchas Informatica and Ascentialwill compete for this spaceagainst the bundledenterprise offerings fromleading RDBMS vendors.

EII technologies providestrong search andfederation base with XMLbackbone, opening up newsources of analyticinformation such as callcenter records, service logs,and web pages.

DataQuality/DataProfiling

TrilliumAscentialFirst LogicEvoke

Initiatives driven by suchthings as SOX will raise theawareness of data quality.

Acquisitions will increase,leaving few pure play DQvendors remaining. Allleading ETL and EIIofferings will provide the DQfunctionality.

12

Data Quality

Much of the data quality market is based on products like First Logic orAscential’s Quality Stage. Is important to note that Oracle’s Warehouse Builder provides strong data quality functionality and, with the release of MicrosoftSQL 2005, we will continue to see strong advances in data quality capabilitiesfrom the RDBMS leaders. Nevertheless, much of the data quality seen to daterevolves around standardization, address matching, house holding, and de-duplication.

In the near term, as more attention is paid to data quality, leading vendors ofthe technology will provide better matching algorithms that learn, as well asinformation enhancement, financial coding structures, and account structures.Driving much of this effort will be the result of SOX and the Basel II accord, inaddition to international pressure for initiatives such as banking anti-moneylaundering.

It is well known, however, that data quality is a subset of tasks and functionstraditionally found in exploratory and predictive data mining technology andunderlying statistics. Consequently, over the long term, as data mining andETL functionality is integrated into the databases, so too will the data qualitytasks. The only pure play, data quality vendors and applications will be thosethat address a particular niche, specifically, information enhancement.

Data Profiling

The current data profiling market is dominated by products like Evoke andProfile Stage from Ascential. While Evoke maintains a market leadershipposition, other products like Profile Stage are better integrated in the ETLprocess. In terms of capabilities, these leading products are similar. They do agood job for single file analysis including simple structure, cardinality, and keyidentification. However, both have limited useful cross table functionality. Thisis indicative of the product class as a whole.

In the short term, data profiling tools will provide better cross table analysiscapabilities. However, not unlike data quality, over the long term, dataprofiling will be commonly offered by your database vendor of choice.

13

OLAP

The term OLAP is generically used throughout the industry to describe thestructure of stored data and the methods we use to access it. The termrepresents several types of varied technology. There are four core adaptationsof OLAP as described by The OLAP Report (www.olapreport.com):

MOLAP -- Multidimensional OLAP refers to proprietary, multidimensionaldatabase OLAP technology. MOLAP is the technology that is built for complex,what-if analysis at the speed–of thought. One word that best describes thisparticular technology is performance. Representative technology includesHyperion Essbase and Microsoft Analytics.

ROLAP -- Relational OLAP is a technology that provides sophisticatedmultidimensional analysis that is performed on open relational databases andother nonproprietary file structures. ROLAP is not bound by the constraints ofother OLAP technology. For example, ROLAP can scale to large data sets in theterabyte range, covering a wide array of informational content. The word thatbest describes ROLAP technology is scalability. Representative technologyincludes MicroStrategy and Microsoft Analytics.

HOLAP -- Hybrid OLAP is an attempt to combine some of the features ofMOLAP and ROLAP technology. The technology provides solid performanceeven while analyzing large data sets, which happen to be the specific strengthsof MOLAP and ROLAP, respectively. However, HOLAP has met with varyingdegrees of success, since it is not a full implementation of all the strengthsindigenous to MOLAP and ROLAP. One word that best describes HOLAPtechnology is compromise. Representative technology includes MicrosoftAnalytics and Hyperion Essbase.

DOLAP -- Desktop OLAP is a technology that is probably the most common inOLAP user communities. DOLAP represents those OLAP tools that areinexpensive and easy to deploy and use. However, the price and ease of usetranslate to limited functionality, especially with regard to the entiredimensional spectrum. A word that best describes this technology isdeployable. Representative technology includes Cognos PowerPlay.

14

Table 5 - OLAP

KeyPlayers

1-2 Year Outlook +2 Years Outlook

Pure OLAP HyperionCognos

The MDD market willcontinue to be dominatedby Hyperion; however, themultidimensional databasewill continue to decline inpopularity except in specificapplication areas such asplanning, budgeting andforecasting. Growingpressure from RDBMSvendors leads to OLAPvendor acquisitions oreliminations.

It will be increasingly difficultto sell MDD technology sincethat capability will beexpected of RDBMS vendors.

MDD technology will be RAM-based due to reducedmemory prices plus greatermemory address range of64bit machines, allowingMDD-like response timesfrom RDBMS technology.Cubes will be built on-the-fly.

RDBMSOLAP

IBMOracleMicrosoft

Businesses increasingly turnto RDBMS vendors formainstream OLAP solutions.Emphasis on techniques togive MDD level performanceto ROLAP queries.

Growing awareness of theadvantages of an RDBMSbased OLAP offeringincluding: integration, lowerTCO and standardization.

Most OLAP storage willbecome transparent to theuser and integrated in theRDBMS of choice.

Real-time capabilities offeredas standard throughvirtualization and policydriven aggregatemanagement. Applicationswritten by specialist vendorsmake up for the lack ofadvanced capabilities.

OLAP is a core component of BI. It affords users the ability to interrogate databy intuitively navigating from summary to detail data. The most significantvendors in this space include:

Microsoft (about 25% market share) Hyperion (about 23% market share) Cognos (about 13% market share) Business Objects (about 7% market share) MicroStrategy (about 6% market share)

15

The OLAP market, however, is changing rapidly. There is a massiveconsolidation of participants; most recently, Hyperion purchased Brio, andBusiness Objects purchased Crystal.

Business Objects and Crystal -- The purchase by Business Objects ofCrystal Decisions is seen mainly as a defensive move, as both were threatenedby the new Microsoft Reporting Services bundled with SQL Server, as well as,the new reporting technology from Cognos. Business Objects paid nearly 57%of its own pre-merger valuation for Crystal in a deal that represents the largestBI consolidation yet. The deal created a company that generates $800mannual revenue, ahead of the $600m by Cognos.

Hyperion and Brio -- The merger between Hyperion and Brio seems to havebeen not only painless, but well received by the market. This merger hasplaced Hyperion as one of the top three vendors for BI products and a directcompetitor with Cognos and Business Objects with revenues of about $600m.

The consolidation will continue in the short term as the largest OLAPcompetitors face an ever increasing challenge from leading RDBMS vendors.The next generation of OLAP will find its data managed by the RDBMS ofchoice. BI architects will find quicker deployment as a result of easiermanagement and administration of multidimensional data. Refer to Table 5.

16

Query and Reporting

Current QR market leaders such as Cognos, Business Objects, and Microsoftare working hard, with frequent updates and new features, to stay ahead inthis very competitive space. Cognos has put a great deal of effort intoproviding a single tool platform with common metadata across its entireproduct line. Although Business Objects had the advantage here, they areincreasing their functionality (as is Cognos) to include more capable securitymodels, richer metadata, and other analysis capabilities outside of reportwriters, including OLAP, data mining, and score carding applications. Microsoft,although late to the dedicated QR market, has put an excellent web reportingsolution in place. Although it is not as mature as Cognos or Business Objects, itis at a price point that the other leading vendors simply can not compete with.

Table 6 - Query & Reporting

KeyPlayers

1-2 Year Outlook +2 Years Outlook

Query &Reporting

CognosBusinessObjectsMicrosoft

Expect furtherconsolidation,commoditization, and pricereductions. Q&R featuresintegrated into the BIplatform, leaving few pureplayers.

Core Q&R featuresintegrated into the RDBMSand BI platform.

In addition to the feature/function wars that will continue, we will see each ofthe leading vendors provide more open, extensible, shareable platforms in thenear future. One significant initiative will be the integration of metadata withother tools such as ETL. In this timeframe, this integration will be mostlystructural, but will allow enterprises a single point of analysis for activities suchas change impact analysis.

17

Dashboards

Currently, the direction of dashboard products is highly influenced by conceptssuch as standardization, Six Sigma, Total Quality Management, PerformanceManagement, Balance Score, and event-based business activity monitoring(BAM). While some dashboard competitors focus on vertical applications,others have tuned their dashboard technology to address real-time analytics(zero-latency).

Many dashboard offerings provide pre-defined templates or entire solutionsuites. For example, MicroStrategy provides literally hundreds of pre-definedkey performance indicators (KPI), each addressing specific businessrequirements. Another example is Panorama Business Views; they offerdashboard solution suites for vertical applications such as, Human Resourcesand Revenue Assurance for the Telecommunication Industry.

While some dashboard competitors emphasize solutions and templates forvertical industries and applications, others have focused on processing issuessuch as strategies for sense-and-respond. This involves the capture andprocessing of complex events, combined with historical data and reported touser communities through BAM dashboards. Leading competitors in this spaceinclude iSpheres and Metatomix.

Table 7 - Dashboards

KeyPlayers

1-2 Year Outlook +2 Years Outlook

Dashboards SiebelHyperionPanaromaMetatomixiSphereIterationBusinessObjectsCognos

HeadsUp:Hyperion

The use of dashboards willcontinue to expand; thus,creating a significantmarket opportunity for newentrants and innovation.

Continued evolutionbeyond widgets, leading toawareness that anunderlying infrastructure isnecessary to pull togetheractionable insight forusers.

This segment will begin tomature, resulting in smallercompetitors being absorbedby the largest BIparticipants, including Q&R,RDBMS, and OLAP vendors.The cost of maintaining andimplementing custom-builtdashboards will drive themarket for packagedsolutions. Pervasivedeployment across all levelsof business.

18

Still there are other vendors who have carved out an important niche byfocusing on real-time analytics. For example, Iteration Software addresses thegap between traditional BI dashboards and their real-time dashboard offeringthat reports KPIs. Specifically, Iterations’ twinkling dashboard differentiates it from other competitors.

Aside from focusing on specific solutions and process issues, each of thesecompanies attempts to enhance the metadata captured and reported by theirdashboard technology. In the case of the SAS, dashboards are a small part oftheir strategic performance management approach that includes not onlydashboards but also other BI techniques and technologies, from data mining toOLAP. Or, with regard to MicroStrategy, they emphasize the ability to readilyflow between Balance Score Cards and dashboards.

Dashboards have the ability to be the standard for many business users in thenear term. Refer to Table 7. We will see dashboards combining much moremetadata, allowing users to search for information very easily. As well, we willsee scorecards being more dynamic and used for interactive business planningactivities.

In the future, dashboard vendors will include rules and reasoning engines intothese products so they will have the ability to “advise” their users. And,integration of external “web” information will become available to compare andcontrast internal information to world information.

19

Data Mining

The data mining segment fosters a cottage industry. There are very few largeparticipants such as SAS, SPSS, and IBM. Instead, this segment is pepperedwith literally hundreds of small, boutique participants. The two leading causesfor the numerous data mining companies are:

Most data mining algorithms are public domain Data mining applications require specificity

Not yet widely accepted by user communities and IT, the need for data mininghas reached broad awareness from vendors. This is due to the simple fact thatcurrent BI solutions must grapple with the rising flood of data, both in terms ofthe number of records as well as their size, and do so more rapidly. Althoughthe rising volume and size of data sets definitely compromises theeffectiveness of many BI-centric tools, it is a characteristic mining is uniquelysuited to address.

Data mining technology feasts on the detail data and can crunch throughmountains of it. This is the environment in which mining thrives, and is one ofthe critical reasons mining will grow as an integral part of any BI process.

Epitomizing the acceptance and integration of data mining, at least at thevendor level, is the Predictive Model Markup Language (PMML). This languageallows data mining experts to create mining models in one environment, andexecute those models in production in another environment. For example, youcan create a SAS mining model using SAS Enterprise Miner, export it to PMML,import that PMML into DB2, and execute the model.

It is the capability afforded by PMML as well as the aggressive integration ofdata mining functionality in all leading RDBMS vendors (IBM, Oracle andMicrosoft), that ensures the acceptance of data mining functionality in BIapplications in the near term.

In the long term, mining will simply be another BI tool. Refer to table 8.

20

Table 8 - Data Mining

KeyPlayers

1-2 Year Outlook +2 Years Outlook

DataMining

SASSPSSIBMOracle

Growing awareness of datamining value will continue.Leading RDBMS and BIvendors will continueintegrating data miningfeatures and functionalityinto their products.

Data mining will becomeanother BI tool bundled inthe RDBMS of choice. Andeven though general datamining will remain theprovince of the expert,vendors will deliverspecialized componentsembedded in a wide-range ofapplications. Operationalanalytics becomes the newmining playground thatresults in the rebirth of datamining allied to rulesengines, correlation engines,and alerting capabilities.

21

Metadata

The current metadata market has two distinct groups of vendors. There are thegeneric metadata or enterprise repository vendors such as CA, and then thereare the ETL tool vendors. The enterprise vendors are suitable for large scaleenterprise repository initiatives where organizations are committed toproviding both the infrastructure as well as the resources (money, manpower)to make it work over extended implementation lifecycles. The ETL vendormetadata solutions are mainly focused on the ETL effort itself, augmenting theETL capabilities with metadata functionality such as data lineage and impactanalysis. Ascential and Informatica lead the ETL industry with the bestmetadata solution right now, at least from the perspective of customer tractionand ETL support.

In the near future, we see the pure play ETL vendors really opening theirmetadata solution to other vendors and tools. The repositories will become farmore extensible and able to deal with solution architectures such as distributedrepositories. This affords the possibility for multi-vendor, integrated metadatarepositories, allowing independent tools to extend each others’repositories andto share their information. A Google like search facility will provide both themetadata and the actual data in a single interface.

Table 9 - Metadata

KeyPlayers

1-2 Year Outlook +2 Years Outlook

Metadata PlatinumInformaticaAscential

Data Integration vendorswill continue to evolve themetadata capabilities.RDBMS vendors will offerintegrated metadatamanagement.

Central metadatarepository products suchas CA will see minimalopportunity.

Inter-vendor sharing ofmetadata will become thecommon approach asopposed to an enterprise-wide, hub-n-spokearchitecture.

Metadata is cornerstone tosuccessful BPM;consequently, as the BPMmarket grows so too will theeffort for improvedmetadata management.

22

Appendix A –Glossary

Table 10 - Glossary

Architecture Unless stated otherwise, the term architecture represents awell-defined and formally documented combination ofsoftware, hardware, and implementation guidelines orstrategies.

Business Intelligence(BI)

BI is the gathering, managing, and analysis of data that istransformed into actionable information.

COTS Commercial Off-the-shelf software.

Data Warehouse (DW) The persistent data structures and related processes tosupport the gathering, integration, cleansing, andmanagement of data for the expressed purpose of strategicanalysis.

Enterprise ApplicationIntegrator (EAI)

An application whose foundation is based on messagebrokerage technology.

Extraction,Transformation andLoading (ETL)

Extraction, Transformation, and Loading are a standardcomponent of virtually all warehouse iterations. It is duringthis process that data is acquired from source systems,converted into relevant data for subsequent analysis, andthen loaded into the data warehouse target structures.

HOLAP Hybrid OLAP is an attempt to combine some of the featuresof MOLAP and ROLAP technology. One word that bestdescribes HOLAP technology is compromise.

Metadata Data about data. Metadata describes characteristics of data,such as size of field, type of data (numeric, character),whether the field is optional or must have a value, validvalues, etc.

Message Broker Is a technology that replaces point-to-point applicationintegration with a standardized system of messaging thattransports data between applications, building a single-point integration environment.

Mining Model Mining models are predefined statistical functions and datapreprocessing required for a particular discovery effort.

MOLAP Multi-dimensional OLAP refers to proprietary,multidimensional database technology. MOLAP is built forcomplex, what-if analysis at the speed-of-thought. Oneword that best describes MOLAP is performance.

23

Normalization Represents business data requirements in its simplest form.Normalization in this context refers to the elimination ofredundancy and inconsistent dependency.

OLAP Online Analytical Processing is a generic term used todescribe multidimensional data and related analysis. OLAPsupports techniques such as drill-down, roll-up, and datapivot. There are three core adaptations of OLAP, includingMOLAP, ROLAP, and HOLAP

ROLAP Relational OLAP is a technology that affords the samefunctionality of MOLAP except the source data is stored in arelational database. ROLAP can scale to large data sets inthe terabyte range, covering a wide array of informationcontent. One word that best describes ROLAP is scalability.

TCO Total cost of ownership.

24

Appendix B –References

Advanced Visual Systems, Inc. Revolutionizing the Display of Business Data.Agosta, Lou. Using an Economic Analysis to Drive Data Warehouse Success. ForresterResearch, Inc., September 15, 2004.Business Week. Getting A Grip On Grid Computing. October 18, 2004.Butte, Brian. “Solving the data warehouse dilemma with GRID technology.” IBM Corporation August 2004.Card, Stuart K., et al. Readings in Information Visualization: Using Vision to Think.Morgan Kaufmann Publishers, 1999.Dresner, H. Predicts 2004: Business Intelligence Technology Directions. Gartner, Inc.,December 5, 2003Friedman, T. Magic Quadrant for ETL –2H04. Gartner, Inc., September 8, 2004.Gassman, B. Data Quality Problems Inhibit BAM Initiatives. Gartner, Inc., July 3,2003.Gonzales, Michael. “More Than Pie Charts.” Intelligent Enterprise November 14, 2004.---. Spatial Business Intelligence. Integeo Pty Ltd., November 2004.---. “The No-Sacrifice Affordable Data Warehouse APP.” Intelligent Enterprise October 30, 2004.---. “Data Quality Discipline.” Intelligent Enterprise October 16, 2004.---. “The SQL Language of OLAP.” Intelligent Enterprise September 18, 2004---. “BI On A Budget.” Intelligent Enterprise April 17, 2004.---. “Breaking Out of the Warehouse.” Intelligent Enterprise September 17, 2003.---. “Enterprise Data Quality For Business Intelligence.” Teradata MagazineOctober 2003.---. “The OLAP-Aware Database.” DB2 Magazine Quarter 2, 2003.---. “Data mining: Can you dig it?” Teradata Magazine Quarter 2, 2003.---. IBM Data Warehousing. Wiley Publishing Inc., 2003.Hostmann, B. Microsoft’s BI Strategy Is a Work in Progress. Gartner, Inc., May 19,2004.Howson, Cindi. M5A –Evaluating Business Intelligence Toolset., The Data WarehouseInstitute, August 2004.MySQL AB. A Guide to Lower Database TCO. December 23, 2003.Rohm, Howard. A Balancing Act. Perform –Performance Measurement in Action,Volume 2, Issue 2.Rubin, Jon. “The DB2 Framework for Business Intelligence.” IBM Corporation May 2003.Tiedrich, Alan. Cognos Series 7 Business Intelligence. Gartner, Inc., Feburary 10,2004.

---. Business Intelligence Tools: Perspective. Gartner, Inc., June 19, 2003.---. Cool Vendors in BI, BAM and Data Warehousing. Gartner, Inc., March 25,

2004.Woodbury, Henry. Why Your Ideas Need Visual Explanation, Dynamic Diagrams Inc.,October 2003.