Review—Data Modeling for the Business

Oil IT Journal reviews ‘Data Modeling for the Business’ by Steve Hoberman et al. The book outlines a new approach to data modeling and includes a chapter on BP’s enterprise architecture.

Someone once said the ideal number of data modelers is one*. The book ‘Data Modeling for the Business**’ (DMFTB) takes practically the opposite approach, advocating a series of corporate Rolfing sessions and pizza parties to thrash out what should be modeled, how, and for how long information should be retained. If the single modeler approach presupposes a domain specialist who knows all, Hoberman’s is rather of journeymen data modelers, perhaps without deep domain knowledge, who can extract all the information required from other stakeholders. The thrust of DMFTB is communication and debate with non specialists. This can be rather labored—as in the first chapter which plods through the analogy of a data model and a blueprint for a house.

Those expecting technology insights and a discussion of tools will be disappointed. We learn from the frontispiece that the graphical models in the text were created with CA’s ERwin tool. But the book does not really connect with technology. The subtitle of ‘aligning business with IT using high level data models’ says it all.’ This discussion is far removed from databases and SQL and focuses on a bird’s-eye view of the enterprise rather than on implementation.

There are ‘traditionally’ four levels of models—very high, high, logical and physical. High level models communicate core data concepts like ‘customer,’ ‘order,’ ‘engineering,’ ‘sales.’ Even the ‘logical’ is model is ‘a graphical representation of [...] everything needed to run the business.’ All of which is a far cry from the Express logical model of Epicentre or ISO 15926!

The body of DMFTN is concerned with business, rather than technical data, examining in depth how for instance the concept of ‘customer’ can be implemented in ‘hundreds of database tables on a variety of platforms.’ Business requirements may mean changing definitions of key concepts like customer. These start at the high level, and ripple down through the model layers. Modelers can then perform impact analysis to see ‘what changes are required at the logical and physical levels.’ Although how such changes are effected across ‘hundreds’ of databases including pre-packaged behemoths like SAP is glossed over.

Of particular interest is a chapter on data modeling in an international energy company by BP’s Mona Pomraning. Here an enterprise architecture initiative set out with a vision of a ‘shared corporate data asset that is easily accessible.’ Amusingly, half way through their work, the team found that there was another initiative working on master data management whose goal was also ‘a single version of the truth.’ Such is the nature of the large decentralized beast! BP’s modelers leveraged industry data models including PPDM, PSDM (ESRI), MIMOSA and PRODML—although exactly how these different circles were squared is not explained!

Despite its technical weakness, DMFTB makes an interesting and perhaps inspiring read for technologists who are trying to engage with their fellow stakeholders.

* links/1003_4.

** A handbook for aligning the business with IT using high level data models. Hoberman et al. Technics Publications 2009. ISBN 9780977140077.

Click here to comment on this article

Click here to view this article in context on a desktop

© Oil IT Journal - all rights reserved.