Repositories: Bottoms Up

Say the word Repository, especially Enterprise Wide, and most IT minds quickly do a word game and you have “Suppository”. And an apt picture of the many excellent tools that have gotten shipwrecked in this field. Think Unisys, Platinum, Microsoft, and IBM for major names in reverse alphabetical order – and roughy effort and technical competence/completeness of the offerings.

Now despite the fact that this developer would dearly like to be able to go to one central place and find all the development resources available to him: Web Services, databases, application servers, EDI connections, CORBA links, Flash components, Java Frameworks, ASP Templates, XQuery schemas, etc, etc. And this list of resources or developers compendium would also direct me to the correct connectors, methods, and APIs needed to access these resources. And this compendium would let me know what permissions I had or that were required to access these resources(not so dangerous, given the right security context). And this compendium would point me to how-to resources like documentation, API descriptors, and code templates and samples which actually utilized aspects of the API. And this compendium would … but I am not going to ask for such a system even though components of it are being developed through database and XML schemas, application server configuration and management files, Visual Studio/Eclipse/Sun Studio team developer workplaces, UML and requirements definition systems, etc.

I am simply not going to ask for anything that resembles a Universal Suppository …. uhh Repository. Why ?

Because low and behold looks what is happening on the IT scene – repositories and other metadata resources are being built from the bottom up as organization-wide data integration starts to take hold. BI systems have to have good cross database, data warehouse, and datamart metadata repositories. Add BAM and BPM and suddenly semi-structured documents and resources have to be added to the list. But also ECM-Enterprise Content Management systems have to have group and organization wide filelists , dictionaries, and resource databases. Ditto for ERP, SCM, CRM – they are all driven on increasingly wider viewing metadata or repositories. In effect what was missing in the universal repositories from Unisys-et-al was the next two or three tiers down of metadata resources. Now those resources are starting to build up. In addition the XMI and other XML metadata exchange routines/standads are are starting to emerge. Instead of moving data, more systems are moving data about data.

However, another realization has kept across the IT scene. IT has to be agile. It has to be able to respond very quickly to changed cirumstances, new groupings, coalitions, mergers, and divestments. In effect, most of the basic flows of information and data across an organization are subject to some change year after year. But also the emerging technology itself also creates new economics, new opportunities or simply the need to meet competitive thrusts with better IT methods. Finally, against this backdrop markets and businesses are moving themselves in the array of products and services that have become accepted by consumers in those markets. All three of these trends mean changes to systems. But in turn these changes present both technical and social/economic/group political dynamics and potential problems. We will only consider the technical problems.

More and more system architects, CIOs, system modelers are becoming convinced to be agile means to be able to manage change in systems effectively and efficiently. Not to try to forestall or divert change but to shape and direct it through cross platform, open, and standardized processes and methodologies. Now there is one vendor that would beg to differ with this asessment but the Microsoft Pipedream will be subject of another note. Meanwhile, the trick is to match these agile processes and methods to the resources and systems available. Well well well.

What are those resources and systems available? How can we discover and utilize them most effectively in the organization? These are universal metadata and repository type questions. But unlike before three conditions now exist:
1)the bottom up repositories and metadata stores have started to appear naturally all over organizations;
2)the XML, n-tier distributed processing, and RDBMS tools to manage and exchange info between universal and subsidiary repositories and meta data resources are flourishing;
3)Systems architects and designers are more inclined to acknowledge conditions 1) and 2) exist and independently are looking for automated solutions to help in handling a whole range of problems in being more organizationally agile. These are methods such as ESB, SOA, MDA, ERP, and others.
The net result, near-universal repositories will start to spring up and then an Open Source grand-unifying implementation will probably carry the day. All I can say is bottoms-up to this proposition.

(c)JBSurveyer 2006