Where can we find a top class CMDB?

[tralfamadorian]

Hi there

We're currently running our 3rd attempt of designing a CMDB. The first two failed miserably, and it seems unlikely that we're going to make it this time.

Why? Because the designers ask everybody what they'd like to register in the CMDB, i.e. what attributes should be registered for each individual CI. As "everybody" includes Network architects, z/OS systems programmers, Application developers, project managers, etc., the sum of attributes makes the CMDB design a major challenge, and the population of the CMDB a dead-born project (in my humble opinion).

I'm looking for someone who has faced and overcome this challenge!

How did you manage to build a CMDB that everybody could accept, and what does it look like, i.e. which attributes do you register for each CI?

[rjp]

Well you are not alone at least. And on the bright side you have found three approaches which don't work. Understanding why will help you figure out what will.

Now assuming you want the mother of all CMDBs you might like to consider the Common Information Model (CIM) Standards produced by the Distributed Management Task Force (DMTF). The DMTF is an industry consortium of all the usual suspects, and the standards are open. There you will find exhaustive schemas (in UML and XML) for just about everything you are going to find in an ICT environment. As well as inheritance, relationships, interfaces, and etc., and etc. But if you put the whole schema on a single page you would need an A0 printer and a magnifying glass

Bear in mind these standards / schemas are designed to support interoperability between applications and API frameworks for automating the management of infrastructure. IT's detailed enough to build agents capable of tracking and interacting with system states down to the level of registry keys. (But no it's not Window's specific).

I suspect, however, that you may have a 'scope' issue. The CMDB is not intended to be Deep Thought's memory banks

I like to think of it as the 'map and terrain' problem. Maps contain lots of information, so the natural assumption is that the more information the map has the more useful it will be. This is not the case. Maps are powerful because they abstract and simply the vast ammount of information on the terrain. The London Tube map is (rightly) famous for being a revolutionary breakthrough in representing information for just this reason. The map that contains too much information starts to become not a map, but a simulation, and find your way around the map is exactly as hard as finding your way around the terrain without it.

Anyway, that's editorial. Have a peek at the CIM if you are not already appraised of it. I have some other, considerably simpler material on the CMDB challenge which I have found quite useful. (Though not an actual model of a working CMDB I can hand over) If the CIM is not the kind of thing you are looking for reply to this post and we shall keep the discussion going - of feel free to contact me off forum.

[Nikhil]

Hi,
This certainly the most common problem faced by many during CMDB implementation.
I would like to suggest you to have a look at BMC Atrium CMDB.
Its a product marketed and soled by BMC wich integrated well with the Remedy ITSM products, again owned by Remedy.
BMC Atrium CMDB is a OPPS based architecture and provides a CDM (Common Data Model) which can be made use of Out Of the Box and implemented.
The CDM can also be extended as per requirements.
Also it has been designed based on best practices defined by ITIL for CMDB.
Though expensive, its one of the best options available.
A suggestion out of experience, limit the CMDB sope to start with. Once stable with the limited scope, you can extend it.
That way it will be manageable and you can have control over the entire process.
_________________
Regards,
Nikhil Kulkarni.
Application Analyst
cMango.. The Services Management Company

The taste of low quality lingers long after the satisfaction of low price.

[Guerino1]