In-Depth

Utility Computing Push

HP and IBM flex their utility computing muscles

• By Stephen Swoyer
• 05/08/2003

IBM last week unveiled a raft of new offerings in support of its e-Business On Demand initiative, including a major overhaul to its pricing scheme—dubbed Open Infrastructure Offering (OIO)—which bears a facile resemblance to what Sun announced in February with its Project Orion subscription license program.

This week, HP unveiled Adaptive Enterprise, a combination of new and existing products and services that provide an infrastructure for so-called on demand computing. The goal, said HP CEO Carly Fiorina during the launch festivities, is to help customers wring more efficiency out of their IT infrastructures. To that end, Adaptive Enterprise is both a product- and services-based offering.

Adaptive Enterprise is similar to Big Blue’s e-Business on Demand and Sun’s N1 projects. All three initiatives describe hardware-, software- and services-based infrastructures for utility computing

IBM and Sun have both rushed to champion utility computing, but HP was effectively first out of the gate when it announced its Utility Data Center (UDC) in November 2001—several months before competitor Sun Microsystems Inc., which disclosed details about its N1 utility computing infrastructure in February 2002. IBM unveiled e-Business On Demand in late October of 2002.

Central to the Adaptive Enterprise strategy that HP announced this week is the so-called Darwin Reference Architecture, which the computing giant bills as a standards-based framework that exploits “best-of-breed” technologies to improve integration between business and IT. HP said that Darwin will support platforms such as HP-UX, Windows, and Linux, along with software from partners such as Cisco and BEA Systems Inc.

Adaptive Enterprise will feature new products—such as an as-yet-unnamed toolkit for integrating heterogeneous operating environments. The new toolkit, which HP says is based on open-standard APIs, will be available later this year. Moreover, networking giant Cisco Systems Inc.—with which HP has partnered in support of utility computing in the past (http://www.esj.com/news/article.asp?EditorialsID=426)—will write a series of APIs in support of UDC.

One area in which HP enjoys a lead over rivals Sun and IBM is in the administration of multi-vendor environments, largely because of the strength of its OpenView systems and network management framework. This week, HP introduced new software, called Virtual Server Environment, designed to help customers manage workloads on HP-UX systems. It is based on a tool that HP developed for HP-UX, called Workload Manager, but will also be available for Windows and Linux environments at a later date.

The new Adaptive Enterprise Strategy will also leverage existing HP technologies and services, such as its Zero Latency Enterprise (ZLE) infrastructure. ZLE describes a fault-tolerant infrastructure for mission critical applications that runs on HP’s Non Stop operating environment and exploits an operational data store to connect business processes in real time. HP first rolled it out in support of telecommunications customers—Sprint was an early adopter—and since then has sought to bring it to a broader range of verticals, including retail and finance.

David Bealby, VP of ZLE telecommunications marketing with HP, says that ZLE is an important aspect of what HP means by the Adaptive Enterprise. “Adaptive Enterprise is a combination of services and solutions and products designed to help our customers include the ability to be more agile. ZLE is a classic, archetypal example of how to do that: At the high end, we’re integrating different data systems together and allowing an enterprise to expect a single view of that information going forward.”

For his part, Tom Steele, a principal network design engineer with Sprint, says that his company has been using ZLE since 1998—when it was first developed by Compaq Computer Corp. Since that time, Steele explains, Sprint’s ZLE infrastructure has grown to support more than 30,000 records per second. “If you look out there at the systems today, there aren’t many that, like ZLE, can do that and protect the data at the same time.”

Steele says that Sprint has watched with interest the utility computing efforts launched by HP, IBM, and Sun. When it determined to enhance its infrastructure by introducing support for operational measurements (which tell it how its network is performing) and alarms (which can detect fraud, among other events), Sprint determined to partner with HP because of its investment in ZLE. “Our current projects are looking at bringing together the local call detail records and operational records, the PCS call detail records and operational records, so that if a customer calls, we can at one point in time tell them exactly what is going on across the network.”

As a high-end component of its Adaptive Enterprise strategy, Bealby acknowledges that HP will market ZLE to a wide range of enterprise customers for which it might be a good fit. For his part, Sprint’s Steele says that ZLE is a relatively painless implementation—it’s designed to accommodate an organization’s existing infrastructure and requires only an HP Non Stop core and an operational data store—and delivers a definite ROI payoff. “There is some change involved with it, obviously, but [it delivers] ROI within two years, it starts tipping so that that platform is actually cheaper.”

New e-Business On Demand offerings from IBM

Capping Big Blue’s e-Business On Demand-related announcements from last week was OIO, a customized pricing scheme that allows customers to make a single monthly payment for IBM hardware, software, and services. It is similar in this respect to Sun’s Project Orion, an initiative that aggregates all of the Unix giant’s disparate software offerings into a single stack. Key to Orion will be a subscription-based licensing plan in which Sun will introduce quarterly updates.

OIO goes Orion one further, of course, in that it encompasses hardware and services. At present, says Eric Stouffer, program director for Tivoli On Demand management solutions for IBM, OIO doesn’t make extensive use of Tivoli’s license monitoring, software asset management, or other technologies. That will change in the future, however: “We have a roadmap to enhance [OIO] so that we better enable billing and pricing metrics. The offering itself is really a way for IBM to bring the pieces together and then build on that so that it’s more cohesive. Tivoli management software will be right in the middle of that.”

For some time now, IBM has installed additional processors in its mainframe, iSeries, and pSeries systems that customers can purchase for use as their capacity requirements increased. Recently, Big Blue enhanced this feature—which it calls Capacity on Demand—to include a “light switch”-type capability, called On/Off Capacity Upgrade on Demand (On/Off CUD). IBM first introduced On/Off CUD with its revamped iSeries systems in January, and this week announced a similar light switch capability for its pSeries Unix servers.

Last week, Big Blue announced a new capacity on demand feature for its Shark storage arrays. IBM indicated that Shark arrays will ship with up to 6.9 TB of reserve capacity that customers can activate when they need it.

IBM is also prepping a technology, called Cayuga, that will allow its WebSphere Application Server to be controlled by the Open Grid Services Architecture (OGSA), software that is used to control highly distributed computational grids. Comments Stouffer: “It combines the optimization and provisioning in this WebSphere environment, so you’re optimizing the execution across multiple systems running in a single WebSphere environment.” Stouffer says that Cayuga will be available as an early beta this summer, but that it’ll eventually be incorporated into WebSphere itself: “The direction is to actually make it part of WebSphere, but it’s early enough in its life to have a focus on it separately.”