“It is like the London tube map,” explains Kosten Metreweli, marketing director for business service management software provider Tideway Systems. “If you saw just the stations with none of the lines that connect them, you wouldn’t know how to get anywhere. That’s what you often have in IT departments: no information on the connections between hardware and applications.”

Without seeing the connections between infrastructure components, it is impossible to predict the effect that changing one component will have on a service. That is well appreciated at global glass maker, Pilkington.

Mike Stops, its IS service centre manager in the EMEA region, says that “the driver for a CMDB is to support and improve the change management process.” Pilkington used IT service management software from Touchpaper to unify the previously separate infrastructures of its regional divisions. “You get significant benefits in viewing the risks of changes through integration to a CMDB.”

A crucial technological development to the process of mapping connections between hardware and applications has been automation. Automatic application mapping was introduced in 2002 by Relicore, a specialist in this area which was subsequently acquired by Symantec, and is now offered by most IT service management vendors (see Forrester Wave).

Discovering the relationships between hardware and applications elevates the CMDB beyond an incomprehensible list of technical specifications, and the fact that the process can now be done quickly and automatically greatly reduces the barriers to achieving ITIL’s IT service management ideal.

And it is not just ITIL that some companies are trying to meet.

As Stephen Ashton, director of global IT business management for investment bank Dresdner Kleinwort Wasserstein (DKW) says, automatic application mapping is a necessity, for companies subjected to the US’s Sarbanes-Oxley governance legislation.

“The burden of proof that Sarbanes-Oxley places on IT departments [in documenting their IT resources] is so great, that you can’t provide the information required without automation,” he says.

Dave Clark, head of software pre-sales for Hewlett-Packard (HP), says the need for automatic infrastructure discovery is particularly acute when an organisation undergoes significant change, such as an acquisition or merger.

“Companies may have informal processes in place to check that everything is working as it should be, but when something like a merger happens and equipment and personnel both change, those informal processes break down,” he explains. Automation lets the IT department quickly re-evaluate its resources in any circumstance.

Another technology that is essential to establishing a meaningful map of the IT infrastructure is federation, a technique that allows items in separate databases to be coupled together without creating one large, unwieldy database.

For Chris Rixon, solutions manager for BMC in EMEA, it is through federation that all data relating to the IT infrastructure, going beyond simple technical specifications, should be associated with the CMDB.

“You need to have the core physical identifiers, so you know what is there and what it is, and that is the CMDB,” he says. “But there is also related information, which could be an incident that happened to an object, or even asset management information. That would be federated to the CMDB, rather than included in it, in order to preserve scalability.”

HP’s Clark says that this model of the infrastructure management data, with CMDB as the central backbone and layers of applied, service-based information emanating from it, is reflected in the companies that do IT service management most successfully.

“We’ve found that they are thinking of the CMDB as the logical data model, with a lot of onion layers around it such as federation tools,” he explains. At the outer layer of the ‘onion’ reside ITIL and business service management tools that translate the data into business process terms, and what binds the layers is federation.

How far do you go?

Despite the promise of automated discovery and federation, building a complete picture of IT services around the CMDB is no simple task. Instead, much like ITIL’s recommendations, it is an end to gradually work towards, and expenditure on that must be rationalised against practical limitations.

“Always implement a CMDB little by little,” advises Thomas Mendel of research group Forrester. “It is very misleading and can be very dangerous to try and do it all at once.”

Tideway’s Metreweli agrees. “When you configure a CMDB in the pure ITIL vision, it is an enormous thing that is all encompassing. It is not something that you can implement in a big bang.”

In implementing a CMDB, there needs to be some prioritisation of IT services that are particularly important to business processes, or have been particularly troublesome.

The focus of IT systems management is often on the big machines in the middle that, if they were to be overloaded, would have the biggest impact upon business processes.

However, as a survey sponsored by Touchpaper found, smaller machines can generate their own share of woes. Out of a hundred CIOs surveyed, 44 listed the constant tinkering with desktop PCs by their users as a significant source of headaches.

Does this mean the configuration of every PC should be logged in the CMDB? Touchpaper’s CEO Graham Ridgway believes that it is unwise to exclude desktop management from central IT service management. “As you move out of the centre of the infrastructure, the significance of each machine to the business decreases. But the number of machines increases rapidly, so the potential impact of failure could be greater.”

The degree to which peripheral technology is covered by the CMDB should be decided on a case by case basis, and according to business drivers, says HP’s Clark. For example, the memory available on a secretary’s PC might not be a salient issue for central IT management, but if an employee uses analytical applications of high business value, the memory on their PC may well be.

Transparent outsourcing

The ideal that companies should be working towards is an understanding of the impact that any change to the IT infrastructure will have on the business, so that such changes can be made more effectively.

An unclear picture of how hardware relates to applications – and so services – is often a barrier to organisational change, says Clark. “With IT service management, a lot of customers come to us with the eventual goal of outsourcing their IT functions. They say ‘give us a simple structure, so we can outsource it’.”

By establishing a master record of hardware availability and application dependencies, which can be viewed from afar, the CMDB enables companies to outsource work to third parties while maintaining a complete and auditable view of service availability.

Formula One racing company Toyota Motorsport, for example, decided to outsource its SAP environment to the automotive giant’s IT arm, Toyota Information Services. Although it was no longer going to directly manage the environment, the team needed up-to-the-second information on the availability of the application, as it impacted on the delivery of vital car parts.

With BMC’s business service management tools pinned around its Atrium CMDB product, Toyota enjoyed the supervision of IT services it needed to operate reliably, as well as the cost benefits of outsourcing the application management.

This value proposition could have wide-reaching implications for the way many companies manage their IT, by enabling them to use outsourcing services without any loss of supervision.

“Rationalising a company’s IT service management forces them to think about what kind of company they want to become,” says HP’s Clark. “And that is part of the value of going through the process.”