Systems management will soon become much more tied to business functions. But data center culture has to change, too, to incorporate and more fully use the tools that are already available and to begin negotiating with end users about the service levels they can realistically expect.
Systems management is going to continue to get easier, especially for helping control the menagerie of multi-vendor systems, applications and piece-parts housed in most of today's data centers.
Over the next few years, more of the tasks needed to help run a data center's operations will become more proactive and less reactive. And increasingly, systems management will be tied to key business functions instead of just specific pieces of IT gear.
Much of this will happen in the context of business service management (BSM) and its related disciplines, designed to promise the most business-critical functions a certain guaranteed level of uptime and availability. For instance, BSM promises that customers will always be able to place orders via the company's Web server -- even if CPU cycles or other resources need to be "borrowed" from the payroll application to do so.
This is opposed to today's service-level agreements (SLAs), already in widespread use, which basically promise that a specific piece of IT gear will be up for a certain amount of the time. So BSM is tied to business processes, SLAs to equipment. It's a big difference.
"A CIO doesn't know or care if a server named Vulcan1 is down," said Dave Hardy, a product manager at BMC Software Inc., a systems management vendor in Houston. "But they do care if the company's ability to process customers' orders stops" because of a down server.
Systems management tools will work behind-the-scenes to monitor, control and -- increasingly -- to fix any hardware or software piece that isn't holding up its end of the BSM bargain.
But all this will happen in small steps, not in one giant upgrade, over the course of at least three years. Progress may be slowed through a combination of technology limitations, data center culture -- and fear.
Many systems administrators see management tools, and BSM specifically, "as a threat," said Rick Sturm, president of Enterprise Management Associates Inc. in Boulder, Colo., "They think that BSM makes things harder and makes things worse." Administrators, and even IT management, often fear that "you put BSM in place and then users will use it as a club to beat them up. And that's not something anyone wants."
Instead, Sturm said, the idea is to use BSM as a means of negotiation and as a way to begin a dialog about users' expectations, as well as what the state of the technology and the current budget will allow. "Eventually, you get into the laws of physics -- if the remote office is in Jakarta, the system hosting the application is in New York and the managing director in Indonesia is demanding sub-second response time, you can talk about the speed of light and how it's a law we can all live
with," he said, only half-jokingly.
But the notion is to understand what users want, and to help educate them about what is realistic and feasible.
In the meantime, there's another shift that will need to happen for this scenario to play out, and that's automating more and more of the traditional systems management realm. Customers need to take full advantage of the capacity-planning and other types of tools already on the market by implementing and really using the functions. That's not always the case, Sturm said.
And the technology needs to catch up, of course. Most of today's tools, for capacity planning and many other system management-related tasks, are point products -- they're focused on monitoring a specific piece of equipment or software. The goal is to help keep the gear functioning and to alert a human being if something goes wrong. Another goal is to help make the network and anything attached to it go as fast as possible.
But the real shift will come as the systems management tools are able to discern which pieces of the IT infrastructure "belong" to which applications, and to make distinctions based on business need. In other words, it may be OK to speed up Application A, and to slow down Application B, because the second is less important than the first -- in the context of the business. That presumes an understanding of which servers, storage infrastructure and software are hosting or belong with various
business processes.
Most of the major systems management vendors -- Hewlett-Packard Co., Sun Microsystems Inc., Tivoli/IBM, Computer Associates International Inc. and others -- are building BSM-based products or have already done so.
In BMC's case, its product suite revolves around building models of business processes, then mapping those processes to the alerts and other existing systems management infrastructure.
Help is also coming from unexpected places. Microsoft, for instance, in addition to providing its usual bevy of tools via Systems Management Server, is building more operations-friendly support into its programming tools. As part of its bid to help customers more easily create Web services that don't clog up the network, Visual Studio 2005 will allow customers to validate any new applications against the company's network topology diagram. This notion is something Microsoft is calling
"designing for operations," said Prashant Sridharan, a senior product manager in Microsoft's developer tools group.
Eventually, observers say, the notion of self-healing systems will wend their way into the data center. But for any of that to really work, it will require a modern systems management infrastructure and BSM as the base.
