By George V. Hulme

Change is risky when it comes to managing business technology. Make a single mistake, and network traffic doesn’t flow, applications get disconnected and transactions die. To keep help desk lights from turning red and to ensure that service level agreements (SLAs) are met, a growing number of companies are turning to what’s known as configuration management databases (CMDBs).

Aligning IT investments with business objectives, long a goal of most IT departments, is a core value delivered by CMDBs. Arising from the popularity of ITIL® (Information Technology Infrastructure Library), which originally inspired the call for configuration management databases, CMDBs are beginning to prove their usefulness at simplifying infrastructure management, resolving problems, lowering the operational risks associated with change and better aligning IT to business services.

“Two years ago, companies were scratching their heads wondering what CMDB meant, and there was a big ‘so what?’ factor,” says Richard Ptak, an analyst at Ptak, Noel & Associates. “But CMDBs have been shown to provide a way for organizations to leverage their existing data scattered throughout the enterprise and quickly understand infrastructure and application dependencies. Once you have that, there are many things you can do.”

The diversity of CMDBs is one big reason why that “so what?” factor has been overcome so quickly. Many companies have already reaped significant benefits from CMDB initiatives. They’re not only gaining superior change-management capabilities, but also attaining the ability to garner accurate system usage and trend analysis — not to mention the ability to predict infrastructure problems that are likely to arise. “CMDBs help you become more effective at problem detection, prediction and analysis,” Ptak says.

That’s been the experience of the Tennessee Valley Authority (TVA), which provides power to roughly 8.5 million residents within the Tennessee Valley area and is the nation’s largest public power company.

TVA began implementing a CMDB a little more than two years ago. “It’s not inventory for inventory’s sake,” says William Brandenburg, interim senior manager, information services, client services at TVA. “[CMDB] is where we collect our list of databases, applications, network gear and servers, along with our service level management information, so we can identify our system dependencies.”

The CMBD proved its worth to TVA quickly. The database provides the ability to rapidly identify the root cause of performance problems. “We now can make sure that we have the right response times for each asset. And our CMDB provides for greater change management, trending and planning capability,” Brandenburg says.

Although root-cause analysis and capacity planning are two of the things CMDBs are geared for, TVA has pushed its configuration management database much further. About a year ago, the power company considered purchasing a specialized tool to track its level of IT compliance with regulations such as Sarbanes-Oxley. “But we realized we already had this tool as part of our culture that was tracking all of our changes and configurations,” says Diane Bunch, senior vice president, information services at TVA. “It’s eliminated the procurement of yet another tracking tool.”

Management software vendors are building or enhancing their offerings to identify, collect and manage configuration information from all of the data stores typically found in the enterprise. In August, CA announced general availability of CA CMDB, which provides a central, federated database that connects — through Universal Federation Adaptors — to disparate data stores. In this way, CA CMDB utilizes the data that already resides in databases throughout the organization, rather than requiring IT managers to maintain duplicate copies of configuration information. Predefined content classes enable the easy importation of configuration information about families of devices such as workstations, servers, routers and other aspects of the infrastructure. “It’s about providing quick value out of the box without customers having to hand-configure these integrations,” says Helge Scheil, Senior Vice President of Development at CA.

Too Much, Too Soon
One CMDB pitfall to avoid is trying to do too much, too fast. Experts advise identifying a lagging business process in the organization and implementing a CMDB to turn that process around. “Look to carve out an area where a big impact can be made,” Scheil says. A good way to begin is by first identifying and mapping IT systems supporting one core business process — for instance online order processing — and then expanding out to map other systems that relate to those systems. Arlen Beylerian, director of product management at CA, adds: “The key is to pick a critical area and build on your successes.”

As companies continue to build and expand their configuration management databases, they’ll provide the foundation that IT operation teams need to more closely, and nimbly, align IT with strategic business initiatives. “For years, the promise has been that IT would make life easier for the operations of the business,” Ptak says. “Through CMDBs, IT will have at its fingertips the current state of all its assets, resources and service levels. And they’ll be more closely aligned with the business than ever before.”

TVA’s Brandenburg agrees. “I have to step back to realize how much it’s changed our culture,” he says. “Not having a CMDB is like flying blind. I just can’t imagine trying to do work — or plan work — without it.”

With mainframes approaching their 43rd birthday, these high-end servers are still in the prime of their careers. Distributed computing began its rise to prominence in the early 1990s, and mainframe shipments today remain strong—as does the fundamental measurement of mainframe capacity, millions of instructions per second (MIPS). For years, the number of production MIPS has been growing steadily, from 15 percent to 20 percent annually.

“Companies are using mainframes for the same mission-critical, highly available applications as always,” Gartner analyst Mike Chuba says. “But specialty engines—whether for Linux or Java workloads—are driving a lot of the new mainframe MIPS shipments.”

Just ask Richard Russell, director of enterprise application services at Herman Miller Inc. in Zeeland, Mich. About 10 years ago, the $1.7 billion maker of high-end office furniture had planned to move away from its mainframe platforms. But today the company still uses IBM zSeries mainframes as part of a system mix that also includes IBM iSeries (AS/400) midrange systems and servers running on the HP-UX and Sun Solaris platforms. Whether by design, through organic growth, or by inheriting disparate IT systems in acquisitions, many large enterprises like Herman Miller have diverse platforms they must manage. “We’re always going to have a number of technical platforms,” Russell says. “We have a desire to rationalize that as much as possible.”

For Herman Miller, the availability of new IBM workloads, third-party software applications and specialty engines—platforms that provide the ability to develop and run applications less expensively than before — has made a difference. While the company continues to consolidate its platforms wherever efficiencies and improved business processes can be realized, the mainframe is still an important part of the mix.

In fact, due to the mainframe’s steadfast reliability, many enterprises are turning to new mainframe tools and management technologies. These tools and technologies make it possible to bring the decades-worth of data residing in mainframe applications into distributed— and increasingly on-demand and service-oriented —architectures.

Enterprises need to better leverage existing mainframe hardware and application investments, and better align these investments with new business demands. To meet that need, network and system management vendors are enhancing their toolsets to enable companies to manage performance, applications and transactions from their back-end mainframe systems out to their front-end Web applications. There’s little doubt that the move toward service oriented architecture (SOA) is likely to bring as significant a change to the IT industry as client/server architecture did in the previous 15 years. And more enterprises are investigating ways to bring their mainframes and legacy applications along into the new fold.

Centrally Manage
“Companies are investigating ways to connect their legacy applications and expose that data so it can be easily utilized by their distributed platforms and SOA services,” says Richard Ptak, an analyst with Ptak & Noel Associates. “When you look at service-oriented architectures and componentized applications, as well as the dynamic allocation of resources, this is simply part of the technology on mainframe platforms,” Ptak says.

Yet the challenge remains to centrally manage applications, Web services and server platforms to effectively troubleshoot performance issues and ensure that service level agreements (SLAs) are consistently met. When transactions begin to slide, IT operations teams can’t be forced to seek out a dozen separate point management tools in an attempt to spot and rectify problems. They need quick visibility throughout the entire transaction life cycle.

“The big challenge,” explains Vince Re, chief architect at CA, “is to be able to manage all of this stuff side by side: the legacy applications, new work loads, mainframes and distributed platforms. The goal is to make sure that the work of the mainframe and other operations teams is always synchronized.”

Leading network and system management vendors now provide the tools that enable enterprises to do just that. CA’s recent acquisition of Wily Technology allows enhanced application and IT asset management across both mainframe and distributed environments. Its acquisition this past spring of job scheduling software maker Cybermation, with its event driven job scheduling capabilities, rounds out CA’s job scheduling. It’s all part of the company’s move to unify job scheduling tools for both mainframe and distributed architectures. “This kind of integration has been a very important focus for us, ”Re adds. “We provide the ability for companies to manage their entire infrastructure no matter what platforms they choose.”

Ptak says the trend is toward mainframe development and management tools to be integrated with, and managed like, enterprise distributed platforms. “The differences between the two are shrinking,” he says. “The specialized distinctions between distributed and mainframe are being stripped away.”

As development and system management tools melt away much of the complexity associated with languages such as JCL (job control language) and other highly specialized languages leveraged by mainframe operations teams, the distinction between mainframe and distributed environments will continue to blur. “It’s a much different world now,” Ptak adds. “It’s more like operating in a distributed environment in terms of the simplicity of access to applications and information.”

That means enterprises no longer need to treat the mainframe as an isolated back-office beast; its databases, applications and data can be shared like any other federated asset within the environment.

George V. Hulme is a freelance writer based in Minneapolis. He has covered business and technology for nearly 20 years.

ITIL® is a Registered Trademark of the UK Office of Government Commerce.