A quick post to point you to a very insightful article from Rob England, published a few days ago in the itSMF USA newsletter, entitled “CMDB is not a given: Why CMDB is a Bad Idea for 95% of Organizations”
I read it with a smile in my face. About time! CMDBs are being pushed more and more by the big vendors even though success stories are scarce and there have been hundreds of painful failures. Moreover, in the “successful” cases, what´s normally deployed is not a CMDB but something else. Yet, as the vendors call anything “CMDB,” they get away with it.
England is right: if it doesn’t contain services, and the relationships of the entities to the services, it is not a CMDB. Relationships are the hard part, typically the part that is done “at the end.” (It never gets done.)
Now vendors are announcing their automated tools for application dependency discovery. While automated discovery may help (we use it in some projects, too), everybody is overselling it. People are in for a painful awakening when they discover that autodiscovery did not discover what isn’t autodiscoverable. To put it in England´s words: “by definition a service is a business concept, not a physical instance of anything.”
Even Gartner is saying these days that if you try to pass your event data through a CMDB (where typically, you´ve even autodiscovered the size of the socks you´re wearing), everything will crumble and die, something that is not even mentioned in the article. That’s why an agile Real-Time Service Model (RTSM) can be your friend to model services, to perform service-oriented Availability and Incident Management and proper Continual Service Improvement. And remember, you need not meddle with a CMDB to enjoy an RTSM.
Please repeat after me: CMDBs are not a “best practice.” CMDBs are not a “best practice.” CMDBs are not a “best practice.”(If you are curious, read this, or buy England’s book here (digital) or here.)
“The total cost of ownership of a CMDB is immense,” England concludes. I guess that he is trying to tell us that it´s expensive. He says that probably only 5% of all companies can justify one.
It´s good to have fellow ITIL practitioners who can see that the Emperor has no clothes. What do you think?