I chatted with some Business Objects ETL/EIM (Enterprise Information Management) folks today, in a call that was a direct response to what I heard from and posted about Informatica. The core of the Business Objects story can be summarized (albeit brutally!) like this:
- Integrating metadata management and visibility, end-to-end, from data transformation all the way to individual reports, is REALLY IMPORTANT. We do that and our competitors don’t.
- Whaddya mean we’re not high-end? We have the performance advantage, if anything! Don’t hate us just because we’re cheap.
- We actually have superior ease-of-use for complex data transformations. Any area in which our competitors beat us on ease is for low-powered point functionality.
- We have a couple of big wins at big enterprises that aren’t BI customers. This validates our enterprise-strengthness. And overall we get more new accounts every quarter than Informatica.
What I found most interesting (other than the text analytics part of their story), is their emphasis on what may be called data provenance. Compliance is making “one truth” almost as important as BI vendors have long claimed it was. But the more elaborate the data’s journey, the more chances there are it will be damaged on the way. Thus, to have perfect trust in your results, you need visibility all the way back into the source of the data.
Arguably, that’s a compromise position, between those of the relational purists on the one hand and the DBMS2 advocates on the other. Relational purists think everything has to be managed by one rigorous data model, or there’s no way proper constraints can be enforced. DBMS2 advocates say that the best you can or should hope for is loose coupling between systems with rigorously specified public interfaces. If the individual systems can’t be trusted, no scheme will give reliable results. And known inconsistencies can be fixed by something like MDM in the middle, just as it is in any other approach that doesn’t assume enterprises will have single Grand Unified central databases.
I, of course, am among the DBMS2 advocates.