In what is turning into a bare-knuckle contest of blade server vendors, Hewlett-Packard is touting the results of a study it sponsored that claims its systems use 27 percent less power that IBMs blades.
The study compared HPs c-Class blades to IBMs BladeCenter H servers. The study looked at power consumption and external air flow in a number of blades servers and 1U (1.75-inch) rack-mount systems in typical data center situations.
Sine Nomine Associates, of Ashburn, Va., conducted the study.
The study, conducted in mid-January, was a clear counterpunch to a similar report that IBM commissioned and announced Nov. 16, during the unveiling of a new suite of data center technologies.
Both IBM, which is based in Armonk, N.Y., and HP have been battling for dominance in the blade market, where the two companies account for nearly 75 percent of all systems shipped worldwide, according to a Feb. 22 study by research firm Gartner.
A similar study by by IDC showed that HP, of Palo Alto, Calif., overtook IBM in blade shipments in the fourth quarter of 2006, but IBM remained the top seller for the year.
The HP-sponsored study compared the ProLiant BL460c and the HS 21, which both use dual-core Xeon 5140 processors from Intel. The systems were also equipped with up to eight DIMM (dual in-line memory module) slots with 2GB of memory and two, hot-plug hard drives.
The results, said Guy McSwain, director of BladeSystems for HP, showed that the HP blades used 27 percent less power that IBMs systems. The study also showed that HPs blades required 60 percent less air flow.
The IBM study, on the other hand, compared blades that were based on processors from both Intel and Advanced Micro Devices. The results of the Big Blue study showed that IBMs systems were anywhere from 26 to 30 percent more power efficient.
McSwain said that HP was forced to commission its own study as a way to counter the claims found in the IBM-sponsored report.
The HP study, said Dwight Barron, chief technologist for HPs BladeSystems, showed several more data points than the IBM-sponsored report. This fact, Barron said, will give the companys customers a better side-by-side comparison of the two systems.
Ideally, the IT industry needs to develop uniform standards that all vendors can agree on to accurately measure per watt consumption and give guidance to cooling costs, Barron said.
"I think until something like that is ready, youll be hearing a lot of one-off stories from multiple vendors," Barron said.