A few months ago, Gartner (if you are just landing from another planet, Gartner is probably the most influential IT research firm) released a paper on the state of CMDB deployments. I have some comments to make, hoping that my Gartner Account Manager is not reading.
A forewarning about IT research statistics. Usually:
- a) they are not statistically significant as they survey a very limited number of people,
- b) they probably consist mainly of the very customers of the research firm, which may or not may be companies that are similar to yours, and
- c) you have to understand that few people like to talk about failures (and this is the only reason why the Four Dinosaurs–these overpromising and under-delivering megaframeworks–are still selling monitoring solutions to midsize companies, where they have such a disastrous track record).
And remember that Umberto Eco allegedly said that statistics is the science in which we take two people, where one has eaten two chickens and the other none, and declare that two people ate a chicken each. But I digress.
My point is that Gartner´s paper Hype Cycle for IT Operations Management, 2010 airs a very modest increase in deployments (from 3% to 5% in 2009, and from 5% to 8% in 2010) that were able to model at least three IT services in their CMDBs. This was presented as a positive trait of CMDBs, which are, by the way, also known as financial black holes.
Now, let’s look at this data through a different light.
If only 5% to 8% of deployments were able to reach that point, it means that 92% to 95% failed at the very modest goal of mapping three IT services. And please notice they clearly say “IT services,” not “business services” as in “the services you really need to be interested in.”
Frankly, to map three services is something you can do with an RTSM (Real Time Service Model) in a few weeks. Or less. And you will probably end up with a richer detail of the business context.
I still don’t understand why CMDBs are so enthusiastically recommended. What are they doing in ITIL anyway? ITIL was supposed to be a compendium of best practices, not untested, so called “brilliant” ideas.
The final comment in the same paper is also very revealing about how hard a task it is: “CMDB implementations can take from three to five years to establish, but have no ‘end date’ because they are ongoing projects.” In other words, they are infinite projects that require three to five years to start. Good luck with those.
Don’t get me wrong, I see how a CMDB may make sense for several internal IT processes (such as asset and change management), but there are other projects that can give you tons more rapid value, credibility and visibility in your company.
If you can choose, do start elsewhere. For instance, start creating a roadmap for Visibility and mapping business (not just IT) services into an RTSM.
You’ll thank me later.