Before you invest in a configuration management database (CMDB), slow down; you may not need one. And if you do,...
By submitting your email address, you agree to receive emails regarding relevant topic offers from TechTarget and its partners. You can withdraw your consent at any time. Contact TechTarget at 275 Grove Street, Newton, MA.
beware vendors selling watered-down versions of the software.
Generally, a CMDB is an inventory database, but it also conveys dependencies among the various components in a data center and depicts the relationships that make up the IT services delivered, according to Ronni Colville, research director at Gartner Inc.
Still, what constitutes a CMDB is a matter of debate.
Will the real CMDB vendor please stand up?
Richard Ptak, an analyst with Ptak, Noel & Associates, said that vendors are mislabeling their databases as CMDBs.
"Among the best-proven CMDB products are those available from BMC and CA," Ptak said. "IBM's, which has been available for almost a year, appears to do a reasonable job. HP was struggling with multiple, overlapping and lagging CMDB technology. Since their takeover of Mercury Interactive [MI], they have settled on what was MI's CMDB, which they claim will work with all of the existing HP service desk technologies (CMDBs)."
Meanwhile, software vendors are benefiting from the IT Infrastructure Library (ITIL) hype, selling products that don't necessarily encompass all the functions of a true CMDB, whose functionality includes both configuration management and change management, said Colville.
Because CMDB technology is still immature, users should evaluate vendors carefully and seek user references. Colville says that unless a product includes the technology's essential functions -- such as data integration and federation, relationship mapping, visualization, and reconciliation -- it's not a CMDB.
"If your vendor claims it has a CMDB but it can't do all of these things, it is not a CMDB," Colville said. "Don't let the vendor tell you they have a CMDB when they are trying to sell you a server dependency mapping tool. Push back on them, and eventually they cry uncle."
Some CMDB wannabes include service desk vendors like Axios Systems, FrontRange Solutions Inc., Mexon Technology , Infra, Touchpaper and N(i)2. Server provisioning and configuration management companies like Opsware Inc. and Symantec Corp./Veritas are also not CMDBs. Then there are the application development companies like Serena Software and MKS Inc., and performance managers like Quest, Interlink Software, ASG, and EMC Corp./nLayers, according to Colville.
Other vendors adhere to ITIL, leveraging their established service desk and IT asset management tools to act as a CMDB. Some offer integrated problem, incident and change management but aren't able to federate multiple data sources, for instance.
Colville predicts that with the CMDB market in flux given frequent vendor mergers and acquisitions, standardization in the CMDB space won't occur until at least 2012.
Until then, the CMDB landscape includes HP's uCMDB, IBM's CCMDB, CA's CMDB, BMC's Atrium and Managed Objects' CMDB360.
And vendors that offer true CMDB products build them to be compatible with their own hardware, not that of other vendors. "The products out there are nowhere near compatible with a multi-vendor environment, which leads to vendor lock-in," Colville said.
Do you really need a CMDB?
When polled, some 40% of 150 attendees at Gartner's IT Infrastructure, Operations & Management Summit in Orlando, Fla., said they are in the process of implementing a CMDB. About 15% plan to install one within six months, 18% will do so by 2008, and another 18% said they plan to implement a CMDB project sometime in the future.
But despite what may sound like some considerable rates of CMDB adoption, don't assume the technology is right for you. Gartner's Colville advises clients not to rush to buy a configuration management database; indeed, not every data center needs one. If you're considering a CMDB, here are some key questions to determine whether to make the leap:
- Do you know which applications reside in your infrastructure?
- Do you have fewer than five configuration discovery management data sources?
- Do you know the business impact of a change to a component in the infrastructure or applications? What percentage of your changes are unplanned?
- Do you know the components of a business service such that, in the event of an outage, you can identify the likely causes?
If you've answered yes to all these questions, you may not need a CMDB; instead your best option may be some sort of mapping tool, Colville said.
On the other hand, if a CMDB is essential to your data center, develop a strategy for implementation. Without one, the likelihood of failure is significant. Colville predicts that 75% of users who adopt CMDB will fail, making it a costly learning experience.
"You have to make good decisions and take it slow," Colville said. "CMDB coalesces several IT domains into one view, though for many people, it isn't doing that."The CMDB agenda
Some companies have taken on CMDB adoption as part of a larger push to incorporate best practices and make strategic improvements to IT. At Medavie Blue Cross in Moncton, New Brunswick, Director of Technical Services Don McPhee has worked with his IT staff to implement CMDB as part of an Information Technology Infrastructure Library (ITIL) implementation. But the road to best practices has posed difficulties and strained resources.
"Implementing the CMDB has been tough; it's been a slow process and complex -- especially trying to take inventory of everything," McPhee said. "We've spent $100,000 on CMDB."
Even with such bumps in the road to CMDB implementation, there are significant benefits. In addition to being a fundamental component of ITIL, CMDB adoption is driven by the need to meet compliance audits and keep pace with the frequent changes in data centers today, Colville said.
Let us know what you think about the story; email Bridget Botelho, News Writer.
Also, check out our news blog at serverspecs.blogs.techtarget.com.