The inoculated server

Servers have been imbued with so much intelligence that many can diagnose, manage and heal themselves, with the big three hardware makers – HP, IBM and Sun – promising more automation features to come.

Using is believing Two more for self-management Components of a self-sufficient server The server foundation

“Server, heal thyself” is the latest mantra of major hardware vendors offering the promise of streamlined IT operations through Lazarus-like “miraculous” technology. IBM is pitching its autonomic computing vision to all who will listen. HP and Sun are following suit with their own variations on utilitarian computing: HP with its utility data center (UDC) and Sun with its wide-reaching N1 initiative (see story, “The evolution of resiliency” ). All three approaches are similar: Create servers, software and related technologies that can heal themselves while interacting intelligently with other networked devices. The result is a wiser, more scalable and cost-effective IT environment.

The time is ripe for such IT wisdom, at least on the server front. According to Forrester Research, Global 3500 firms report server utilization at 60% – meaning $20 billion in new servers was wasted last year.

Moreover, the tough economy has forced companies to squeak out efficiencies everywhere, a reality vendors say they are attempting to address. The theory goes that by making devices more self-sufficient, expensive man-hours can be recaptured from time-consuming, mundane management functions. Take IBM’s recent autonomic computing initiative, a companywide, $10 billion investment in hardware, software, services, and research and development that many say places Big Blue on the forefront of the automation movement.

IBM folded its Project eLiza self-healing server initiative into its larger, autonomic computing scheme and will offer autonomic functions for the eServer line, including the Intel processor-based xSeries, midrange iSeries and Unix pSeries servers. The company also offers autonomic features for the zSeries mainframe servers.

IBM’s eServer products are self-configuring, in that hardware subsystems and resources can configure and reconfigure autonomously at boot time and during run time, according to IBM. Self-configuring servers also add or remove hardware in response to commands from administrators or hardware resource management software.

Further, IBM’s eServer series is self-healing, meaning instant detection of hardware or firmware faults and prompt recovery from them without compromising the operating system and user-level workloads, IBM promises. Self-optimizing features autonomously measure performance and resource usage, adjusting configuration accordingly. Self-protecting features enable the servers to guard against internal and external threats to systems and applications integrity.

In addition, IBM is rolling out new features for its flagship Tivoli network management software that automate tasks across network systems including servers. Tivoli Risk Manager produces periodic “heartbeats” that upstream servers use to verify the operational status of specific server systems, and it monitors for security events across the IT infrastructure and then automates security incident analysis. Risk Manager uses algorithms to correlate security alerts and identify threats to server systems and data, and then it conducts automated responses such as server reconfiguration, security patch deployment and account revocation.

In all, IBM in October laid out plans for boosting its systems management portfolio with 26 autonomic-related offerings, including new identity and storage resource management software. IBM’s Storage Systems Group also announced autonomic features for its Enterprise Storage Server, named Shark. A few products already are available with autonomic features, but most will be upgraded over the next year or more.

Using is believing

Whirlpool already reaps the benefits of the autonomic trend on technical and strategic fronts, says Ronda Kiser, Whirlpool’s senior manager of Midrange & Distributed Operations Services for the company’s IT division, in Benton Harbor, Mich.

“The ability to automatically ‘detect and fix’ a problem at Whirlpool can reduce the amount of time a physical body spends checking logs and digging through the infrastructure,” Kiser says. “The system will do it for us. This should reduce outages, and increase availability and human productivity – a win-win solution.”

The $10 billion appliance manufacturer recently began using supply-chain management software from i2 Technologies on two pSeries Unix servers. This software-autonomic server combination has been instrumental in easing management of Whirlpool’s supply chain, Kiser says.

Whirlpool is a solid IBM shop, running about 500 servers from the Netfinity and RS/6000 lines and a smattering of pSeries Unix servers, including the autonomic 660, 680 and 690. Other than the i2 supply-chain software, Whirlpool runs enterprise application software from SAP and Siebel Systems. It relies on IBM/Tivoli systems management tools.

Whirlpool also has started taking advantage of the IBM xSeries Intel processor-based autonomic servers. The company has more than 300 of these Windows NT servers running IBM Director 3.1, which gives IT central management for systems placed globally. Whirlpool plans to complement that centralized management with Tivoli Distributed Monitoring and perhaps Tivoli Enterprise Console software, for centralized event management, Kiser says. By doing this, she says, Whirlpool has a path toward even more automation for NT, gaining features such as automated help-desk ticket generation. “Self-healing features for NT process failures can be managed automatically by the system vs. by a body,” she says.

Beyond NT servers, Kiser sees the role of automation as valuable for business applications such as SAP. Self-healing can fix the application, or its underlying infrastructure, when problems are detected. She envisions the day when the infrastructure is tied together with end-to-end system management software. That, she says, will “drive down the amount of resources that are required to investigate and resolve problems” Self-healing of servers and other infrastructure components becomes central for IT because “recognizing, evaluating, communicating and healing are the keys to keeping our service-level agreements with the business,” she adds.

Kiser anticipates that such future automation will reduce help-desk calls as self-healing servers will fix systems often before end users experience problems. Or, she says, Whirlpool can automatically message its more than 17,000 worldwide employees that it has identified the problem and is working toward a fix, thereby minimizing calls to the help desk, which Whirlpool outsources to HP.

Two more for self-management

HP also is in the self-managing data center race with its UDC, a line of products that aims to virtualize a company’s data centers into a single pool of resources, including remote locations. Since announcing the UDC product family in November 2001, HP has extended its vision to include server, storage and network offerings. These are integrated, deployed and monitored by intelligent management software.

For example, a Web retailer that needs 25 servers to handle online transactions during the Christmas rush but only five servers during the rest of the year could use UDC to grab capacity from other corporate servers during the holiday season. It could temporarily reallocate capacity from a development environment, a human resources system or an SAP system, says Nick van der Zweep, HP’s UDC director. UDC also provides failover of systems such as firewalls, load balancers and servers running Windows, Linux, HP-UX, Solaris and other operating systems, he says.

HP beefed up its self-managing server muscles with the acquisition of Compaq and its ProLiant line of servers. The ProLiant BL server blade line has intelligent fault-resilient power and integrated, “lights out” remote management features, HP says.

Sun is entering the competition with its N1 initiative. Like its competitors, N1 wants to offer users automation and virtualization – the so-called utility computing environment. To that end, Sun in November made two acquisitions. It bought Terraspring, for its server configuration technology, and Pirus Networks, for its storage switches.

Components of a self-sufficient server Vendors are attempting to make more resilient servers with the following features:

• Self-configuring Servers can reconfigure themselves on the fly to adapt to changing computing needs or upon recognition of new add-in hardware. • Self-optimizing Servers monitor their own performance and spit out suggestions for improvements or automatically self-configure themselves. • Self-protecting Servers monitor themselves for security vulnerabilities and breaches, and take appropriate actions when they detect a threat, such as deploying a patch or revoking an account. • Self-healing Servers perform self-diagnoses and patch themselves or initiate a failover when a subsystem fails. • Automated communications By tying servers into equally intelligent management systems, these independent computers can automatically send out alerts to notify employees that a problem is detected and being addressed, and they can become a virtualized member of a server farm or data center, taking or giving over resources to/from other servers or applications on the fly.

First on the N1 agenda, Sun has added automation features to its server management software, and addressed virtualization. Through software available now users can aggregate servers, storage, even cabling. The Terraspring software creates what Sun calls logical server farms that the software automatically creates and configures. Other NI software reallocates and monitors resources as well. Sun has promised to add service provisioning and policy automation to N1 for delivery later this year.

But Sun isn’t totally ignoring the servers themselves. The Sun Enterprise 10000, for instance, offers systemwide error detection and correction. And, Sun says it will release an N1-enabled blade system sometime during the first quarter.

The server foundation

Despite the current vendor hullabaloo, self-managing servers are a substantial paradigm shift and true automation of them looks to be three to five years out, says John Humphreys, a senior analyst with IDC.

But vendors aren’t the only people that see an automated future. Forrester has dubbed its vision of the automated future as “organic IT.” It calls for an overhaul of server networks, storage, software and processors so that a computing infrastructure automatically shares and manages companies’ IT computing resources. Before these new buzzwords from vendors and analysts appeared, the industry called the concept of the self-sufficient infrastructure the lights-out data center. As Humphreys explains: “Organizations want to integrate their information structure under one roof.”

Whatever the name, servers are the foundation on which the future of the intelligent infrastructure rests.

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