The first thing about this little book is what is ‘viral data?’ Is it good—as in ‘viral marketing?’ Or bad as in ‘swine flu?’ Apparently it is bad. In fact the introduction by IBM-er Tim Davis suggests that 2008 collapse of the worlds financial markets was due to ‘a perfect storm of viral data’ which was news to us. A definition of ‘viral data’ as ‘[producing] undesirable effects by engaging information though a service’ does not get us very far either—although the claim that ‘left unchecked, viral data in a service oriented architecture can reach epidemic proportions in an enterprise’ is a bit more promising—even if it does have a perfume of good ole FUD (fear uncertainty and doubt).
The book is a bit of a ramble, but the theme is that with SOA, bad data gets much better visibility than it does in other systems—so data quality is doubly important.
Fishman claims ‘the overall quality of information in organizations continues to be suspect and poor’ and his book sets out to address ‘the treatment and prevention of harmful data.’ Does it succeed?
Well it is certainly not a recipe book, more a collection of anecdotes and remarks, some of which are relevant to data quality, and many which, while interesting (a rant about PowerPoint, the origins of the Federal Reserve System), are not.
The contents page offers a semblance of structure that is not really borne out in the book. In fact a cover to cover read is necessary to capture all the nuances and anecdotes. Is it worth the effort?
I would say yes if you have the time and are prepared for some frustrating digressions. Fishman’s credentials (IBM program director, DAMA board member) are pretty impeccable. We enjoyed his observation that ‘organizations adopt specific protocols to drive measurable improvement—but suffer from the illusion that they can assure quality by a specification like ISO 9000 [..] this is the ultimate naiveté.’ There is some good advice to data modelers—about the need to get data structures ‘right’ and the risks of subsequent changes to tightly-coupled architectures. The chapters on data governance and reference models were eye openers to us as non practitioners, touching on methodologies like DQA3 and CIDER. But Fishman suffers from a tendency to qualify just about every statement made. Four ‘howevers’ on a page is too much! Spend a few hours reading VD in SOA you will probably want to call in the consultants—which is maybe the object of the exercise!
* Viral Data in SOA, An Enterprise Pandemic, by Neal Fishman, IBM Press/Pearson, ISBN 978-0-13-700180-4.
This article originally appeared in Oil IT Journal 2009 Issue # 9.
For more information or to comment on this topic email here.