As corporations move toward utility computing, where system resources grow and shrink according to business demands, server management tools are evolving to do more than simply monitor and manage isolated network nodes. Analysts say users can expect these management packages to become increasingly automated, better handle security and patch management, and become usable across a variety of different server platforms.
Analysts say they expect traditional server management packages to become integrated more tightly with storage provisioning tools and to develop more application savvy so users can configure servers to meet specific application demands.
"As the move toward automation goes up and the number of servers under management goes up, the management tools have to get more mature," says Jonathan Eunice, principal analyst at Illuminata "Putting more elbow grease into managing servers just seems less and less productive and less acceptable."
In the past, server management tools, such as IBM Director, Compaq/Hewlett-Packard 's Insight Manager, Sun Microsystems' Management Centre and Dell 's OpenManage, have been the less-than-flashy workhorse tools of data centre management. But in the past few years, these types of packages have taken on more sophisticated features such as partitioning capabilities in one management console.
"It used to be people treated server managers to twiddle bits on individual systems," Eunice says. "Now they're doing a lot more to treat the computer as a business asset."
That means users can expect some server management tools — base versions of which typically are shipped with the hardware - to take on features more in line with high-level network management packages such as HP's OpenView, IBM's Tivoli and Computer Associates International 's Unicenter. These packages can acknowledge network components, but are more concerned with overall network performance.
And those packages that don't take on more advanced features will make the nitty-gritty work they do — such as asset management or patch management — happen more smoothly as they feed into the higher-level tools that are used to manage heterogeneous servers, storage and other systems.
"We're getting to the point where a lot of these functions, whether it's remote control or asset management are getting pretty good, but there is still a lot of room to smooth things out," Eunice says. "Patch management is a good example . . . I've heard from two or three vendors about substantial improvements in the precision that you can get (from) those patches and in the security model for handling those patches."
Analysts don't tend to track revenue for low-level hardware management tools because they ship with servers. But the overall infrastructure performance management market, which includes server management, is pegged at about US$5 billion, according to Giga Information Group The firm says BMC Software leads the server management segment, followed by IBM Tivoli, CA and HP.
HP is moving ahead of competitors with its low-level server management tool, analysts say, as it preps an update to Insight Manager that will let users manage Unix, Linux and Windows servers from one console — something that today must be done by hopping from console to console. HP is expected to soon begin shipping the tool, code-named Nimbus — an amalgamation of Insight Manager - which manages its Intel-based ProLiant servers, and Servicecontrol Manager, which handles HP-UX, its version of Unix.
Another vendor focused on managing heterogeneous systems is Amphus , which was founded in 2000 as a blade server company, but now focuses on server management. It recently rolled out an update to its ManageSite server management software to support heterogeneous servers with Intelligent Platform Management Interface so users can provision, monitor and manage any IPMI server from one console. Dell, Intel, HP and NEC developed IPMI to define standard interfaces for monitoring and managing servers.
Standards change everything
Analysts say standards such as IPMI, Web-based Enterprise Management, Common Information Module and SNMP will become more important as vendors recognise the need to manage heterogeneous platforms and systems, and to do it without requiring the time-consuming task of updating agents on individual servers.
"It all comes down to keeping (servers) available, maintained and performing, all this while not requiring an army of administrators that explode our (total cost of ownership)," says Ulrich Seif, CIO at National Semiconductor in Santa Clara.
Seif says he's looking for more than what's offered in server management products today.
"Ideally, we would like deployment functionality to inventory, build, patch and update servers from a central networked source (operating system, software, patches, BIOS, firmware, drivers) with minimal server interruption," he says. "Then we would like a reasonably priced monitoring tool to provide scalable, application-specific — operating system, SQL database, email - monitoring with efficient logging, robust process/performance rules and alerts, and graphic, Web-based reporting."
Rick Beebe, manager of systems and network engineering at Yale University School of Medicine in New Haven, Conn., says one of his biggest server management problems is controlling access to servers. "Managing access is (a) ... big resource consumer," he says.
Controlling access to network resources is an area where analysts expect server management vendors to focus. "There are larger numbers of administrators working on these servers, and (server management tools) are gaining capabilities like hierarchical access so you can have some administrators who are really king of the crowd, while others can only control some servers or some functions," Eunice says.
What users want
James Olson, CIO at Waterbury Hospital in Connecticut, says he's looking for management tools that will let him get a good read on how servers are handling specific applications.
"In healthcare, we need products that can go to the lowest level of network device and ascertain its health," he says. "We need to centralize the running to test applications to ensure devices are operating within specifications. This is very important for (Food and Drug Administration) compliance, and the record keeping is important."
Server management vendors likely will roll out this type of application-aware capability, says Corey Ferengul, vice president and principal analyst of operations strategies at the Meta Group.
"The coolest thing in the works is linking these server management tools with monitoring and having them automatically trigger deployment or redeployment of servers based on monitored needs," he says. "There is quite a bit of logic generally necessary to reprovision a device — one being used for another purpose, for example — but the logic exists today to provision from a pool of waiting devices."
IBM, for instance, is prepping Tivoli Intelligent Orchestration software, which will let users provision servers to automatically respond to application demand and then reprovision themselves to run other applications as demands dictate.
What users should expect, analysts say, is for server management vendors to simplify their tools and make it easier to manage multiple, heterogeneous servers in a way that links them with other important business systems.
Users can expect "more automation, more coverage of software. It will not be enough just to do the server and the operating system, but the tools will have to impact the software layer as well.