What About the Enterprise Repository?

Over theyears, I’ve been watching and waiting for a true enterprise repositorysolution. While a number of companies, including Microsoft, have released enterpriserepository products, to a large extent these have been failed efforts.

It seemsthat the enterprise repository problem is actually multiple problems that havebeen lumped together. First, there is the enterprise application repository.This repository is designed for developers creating customized applications.The problem these tools attempt to solve is application component reuse.Enterprise application repositories from vendors such as Microsoft, AllenSystems Group, Computer Associates, and others, have had relatively limitedsuccess. Their cost, proprietary interfaces, and steep learning curves havemade them more appropriate for large corporations that are willing to make asignificant investment in their deployment.

Enterprisedata repositories have likewise met with limited success. There have beenmultiple instances of vendors with metadata repositories that are intended toprovide a common, uniform infrastructure solution for accessing various datasources. These repositories are attempting to solve two problems. The first isto create a single location where developers can obtain information about thevarious heterogeneous data sources scattered throughout most large enterprises.The goal of this type of repository is to reduce the number of redundantdatabases that result when uncoordinated groups develop applications.

The secondproblem that the enterprise data repository is attempting to solve is to createa uniform interface that allows applications to access data on multiple,heterogeneous platforms, using a single set of database APIs. Thesestandardization efforts have failed, for the most part, for a variety ofreasons. In some instances the vendors trying to put the standards together aredoing it as a countermeasure against a common enemy. IBM, Hewlett-Packard, SunMicrosystems, Oracle, and other vendors often find common cause bandingtogether against Microsoft. Microsoft, on the other hand, has tried to puttogether a comprehensive database API in OLE DB.

Finally,there have been various industry consortium attempts to create a uniform dataaccess API or metadata exchange methodology, such as the Metadata Coalition,which recently announced its merger with the Object Management Group, which isfocused on defining the CORBA specification.

The latesteffort of this type is the relatively new Universal Description, Discovery andIntegration (UDDI) Consortium. UDDI is an initiative to create aplatform-independent open framework. It is intended to support cross-platforme-commerce by publishing a specification for Web-based information registries.UDDI incorporates standards such as HTTP, XML, SOAP, and others. It’s unique inthat it also offers a Business Registry, which enables a company to list adefinition of itself, plus its services. Despite -- or perhaps because of --support from IBM, Ariba, Microsoft, and a host of other companies, I am notterribly sanguine about the prospects of the UDDI to attain critical mass inthe marketplace, which is a requirement for success. The fact that UDDI is notyet fully specified -- it relies on the support of competing companies withtheir own hidden agendas -- further increases my pessimism.

The thirdtype of enterprise repository is the knowledge management repository. Thiscategory of repository is intended to consolidate and categorize knowledge fromvarious sources, such as applications, documents, and databases. One unusualaspect of knowledge management repositories is their focus on capturing“wetware,” that is the knowledge, experience, and best practices contained inthe heads of people who have domain expertise. The more ambitious of theserepositories attempt to get people to enter their expertise into the repositorydirectly. While this is a laudable goal, it often fails unless there is anincentive for the knowledge worker to spend the time and effort organizing andentering this information into the repository. Less ambitious efforts provide ameans for experts to list their areas of expertise. Seekers of knowledge canthen search the repository to identify domain experts and contact thoseindividuals directly for assistance on specific projects.

So, we areleft to contemplate a landscape littered with the efforts of dozens ofcompanies and various industry organizations that have all, to one degree oranother, failed to be adopted as the underlying infrastructure required of anenterprisewide repository strategy. It’s too bad. I guess the bottom line isthis industry is so fragmented and balkanized that there is no single vendor --even Microsoft -- capable of providing a unified, integrated perspective on anysingle organization’s application and database assets. --Robert Craig is vice president of marketing at Viador Inc. (Burlington,Mass.), and a former director at the Hurwitz Group Inc. Contact him at robert.craig@viador.com.