I have been wrapping up the book, and in the closing stages (with some last minute research)I have started to come to some conclusions that are a significant shift from previous positions.
First, I have been hard on Semantic Web based approaches to CMDB standards. I still have profound concerns about the learning curve and cultural issues we are going to run into on this path, but the more I think about the problem the more I am inclined to say that the Semantic Web may be the way to go, with its support for semantic interoperability across complex, decoupled ontologies. More on this later, after the final draft is put to bed (by next Weds...!)
Second, I have been floating the notion of Model Driven Configuration Management for some time now. I am quite gratified that Gartner is now echoing this idea in a recent CMDB research paper. Unfortunately, I have yet to see much OMG-based progress in this area from any of the major vendors. On the other hand, Microsoft's approach to Domain-Specific Languages includes specific, valuable work on just this topic, with the Visual Studio 2005 Deployment Designer. Here's a related presentation (5 mb). (I admit to being behind the times on this one, but I have not been tracking Microsoft much as an ITSM player, because they are so proprietary and ITSM almost always involves cross-platform concerns. They are establishing some leadership here I'm not seeing from the other ALM vendors.) I will dig into this further subsequently...
The book is up to 380 pages, 108,000 words, 150+ diagrams. It's been a journey, especially for a guy with a corporate day job. A Weblog redesign and refresh is one of my first priorities after the book is in, and you'll be hearing a lot more from me on more of the key conclusions that have surfaced during this last consolidation phase.
If you solicited a review copy from me, please send me your comments as soon as possible. At this point I'm looking for high level reactions and significant omissions.
-Charlie
There are two options for building a CMDB. The first is the using MDM from a company like Siperian. The second is the use of Semantic Web Technologies. I lean toward Semantic Web tech for many reasons. here are a few:
1. ability for distributed groups to derive from central schema
2. there is a strong context around instances that does not occur in conventional DB tech.
-Ken
Posted by: Ken Arakelian | May 17, 2007 at 09:28 AM