Wednesday, May 7, 2008

EA knowledge bases need to be active

(prompted by people using old approaches with new tools)

The idea of publishing data in a static format is counter to the efficacy of EA.

What one wants in EA is to allow the ownership of the data to remain where it currently resides (i.e. associated with what business function, role currently owns it) - and to be able to see and analyse aggregations of the data (i.e. the data that represents the enterprise).

People should be able to update the data as a natural by product of their day to day work (function, role etc.). As almost everyone in an organisation is responsible for (i.e. owns) some data - this means that eventually almost everyone should be able to update some aspect of the data i.e. the function of maintaining the data is progressively distributed and decentralised (obviously this requires access management, data integrity policies etc.).

This is true of most other enterprise solutions (i.e. all users can update some things, and many things they just reference)

Actually typically people already do update some aspect of this data (i.e. update of the EA is already distributed and decentralised) - usually in a plethora of disconnected and unstructured documents (that suit each individual) but nothing for the enterprise (or the others in the enterprise) . Much of it would be in documents such as: business cases (applications, systems, business processes/products and services, business goals and strategies etc.), operational and procedural documents (holding roles, business processes, rules, products and services etc.), systems diagrams (applications, technology infrastructures), definitions of bodies of work e.g. project charters, statements of work, technical design documents, SLAs, BCP and DR documents etc.

The idea of publishing implicitly involves an unnecessary and function i.e. someone/something has to "publish" it. Where as in fact it should live (active data). The old model is like someone writing the encyclopaedia Britannica and "publishing" it. This just doesn't typically work for EA (it involves rework, it is too slow, the owner can quickly update or correct data, it is counter to the actual process of operating the enterprise).

Having said this in the initial stages of establishing an EA (perhaps much of the 1st year). It is reasonable that some people with a particular interest in a set of data central to many things e.g. business functions, application services etc. may establish a base of data, and some ways of organising the data, to make it easier for everyone else to collaborate.

Using an active EA as if it was is as static EA - is like using a nail gun as if it were a hammer (and then wondering why it takes so long to nail in the nails, and noting that a nail gun costs more than a hammer).

No comments: