• United States
by Mary Ryan-Garcia

Clash of the server titans

Dec 23, 20026 mins
Computers and PeripheralsIBMNetworking

IBM, HP and Sun are sparring over strategies for building self-healing, self-managing server networks.

A battle royale is brewing among the server titans. IBM, Hewlett-Packard and Sun are busily mapping out strategies each hopes will give it the edge in the envisioned new world where self-healing, self-managing servers rule the network.

IBM has been the most vocal of late, aggressively promoting its autonomic computing concept, which includes self-healing, self-managing servers. Alan Ganek, vice president of IBM’s new autonomic computing division, paints a compelling picture of the networked world in which ultrasmart servers operate.

When outfitted with special software such as IBM’s Electronic Service Agent, he says, enterprise servers that are having problems automatically would send information about symptoms to a computer at IBM. Tapping into a comprehensive problem database, the computer would analyze the problem report and initiate corrective actions. It might send an electronic response to fix the problem itself, place a call to the enterprise’s administrative staff with recommend fixes or dispatch an IBM engineer to perform repairs.

“The result is simple — problems are fixed quickly, sometimes before you even knew you had a problem; your infrastructure becomes more resilient and downtime is minimized, thereby lowering your maintenance costs,” Ganek says.

A ubiquitous problem

Though generating a lot of buzz recently, the notion of self-managing servers has been around for several years within the high-end server market, says Vernon Turner, a group vice president at IDC. The battle is heating up now, he says, because of one simple fact: “As servers become more and more of the industry standard or commodity devices, it is harder to differentiate each vendor from another.”

Couple this with the reality that enterprise infrastructures commonly consist of thousands of servers — servers that users are demanding be easier to manage, always available and run at top capacity — and vendors have no choice but to rejuvenate their product lines, Turner adds.

Richard Fichera, a vice president with Giga Information Group, attributes the intensifying efforts around self-managing servers to an identical set of customer problems each vendor faces: stranded capacity, complex installations, difficulty of provisioning new applications and services quickly, and the inability to manage server networks on a service vs. an element basis. “Server management is a ubiquitous problem across all installed bases,” he says.

And as server management issues escalate within enterprise organizations, talk of autonomic computing and utility data centers has been a real attention-getter among users. “The ability for anything to possibly manage itself is something that we would like to see, and it’s currently on our radar screen,” says Aaron Merriam, network operating system/messaging specialist at Hannaford Brothers, a regional supermarket operator in Scarborough, Maine, that has about 400 servers today.

One potential promise Merriam sees in self-healing, self-managing servers is cost-containment. With smarter servers, Hannaford could potentially maintain IT staffing levels while increasing the number of servers. Another potential boon from self-healing, self-managing servers would be the opportunity to give IT staff the time to focus on the future rather than always operating in maintenance mode, Merriam says.

The battle plans

IBM realized autonomic computing would be an industry “grand challenge” more than a year ago, when it launched Project eLiza, the code name for IBM Server Group’s efforts to deliver autonomic computing capabilities, Ganek says. The features, many of them originated for IBM mainframes, will now be deployed across IBM’s eServer family.

While IBM has garnered much attention of late because of its autonomic computing splash, HP  actually has an 18-month lead over its competitors in self-managed servers with the Utility Data Center (UDC) line it introduced in November 2001, Gartner says. UDC products, which let server administrators provision services on demand, include linked server, storage and network operations offerings that are integrated and deployed by intelligent management software. The HP UDC allows remote server management and includes management racks that automatically discover the topology of a user environment. It also offers an Integrated Service Management rack, capacity planning and optimization software, and storage arrays. With a UDC infrastructure, once servers are wired together, configuration and feature changes are handled via software updates.

Sun, while avoiding the self-managing buzzword, is pushing self-monitoring services heavily as part of its N1 open architecture. Sun previewed the multiphase N1 scheme in February and unveiled the formal road map in September.

Phase 1 calls for the virtualization of all network elements – servers, storage and cabling included – for easier and more cost-efficient management than is possible today. Through new system software, available now, users would be able to transform individual computers, network and storage systems into an aggregated pool of resources. The software will allocate, monitor and meter resource usage, in part for service-level management and accounting purposes. The software also will be responsible for life cycle management of services, performance assurance and security.

Subsequent phases, for 2003 and beyond, will enable service provisioning and policy automation. Overall, N1 enhancements will be found throughout Sun’s product line, from operating and file systems to servers and development tools. Yousef Khalidi, N1’s chief architect, explains the main difference between managing servers and other network resources today and tomorrow, under Sun’s virtualized architecture: “Today, IT administration has to deal with every service component individually. With N1, in a ‘virtualized’ environment, much of that low-level complexity gets managed for you so that you don’t have to manually perform all of those tedious, time-consuming and error-prone low-level tasks.”

To be victorious

Today’s IT managers really can benefit from having an enterprise environment capable of being treated as a single entity, IDC’s Turner agrees. “The major server vendors are heading in the right directions,” he says, noting that IBM, HP and Sun each have unique leads in particular parts of the race, while none of them is showing a clear market lead.

Victory in this server war, Turner says, will go to the vendor that achieves the best integration among the chip architecture, the I/O configuration, the operating system and the suite of applications taking advantage of these, resulting in a coherent user solution.

Ryan-Garcia is a freelance writer in Coram, N.Y. She can be reached at



Leading vendors are trying to best each other with wide-ranging strategies for building advanced intelligence into their servers making them capable of self-healing and self-management.


IBM, with its autonomic computing; HP, which touts the Utility Data Center; and Sun, with its N1 initiative.


Analysts say the server vendor that achieves the tightest integration among the chip architecture, the I/O configuration, the operating system and the suite of applications taking advantage of these will be the most successful.

User impact:
These advanced servers will give enterprise users the ability to better use server capacity and to grow the number of servers without adding staff, while reducing the time spent on administrative minutiae.