IBM Sharpens Blade Servers

IBM embeds SAN, Fibre Channel connectivity into its line of BladeCenter ultradense servers.

NEW YORK—IBM is adding another layer to the shared infrastructure of its blade server architecture.

The Armonk, N.Y., company on Wednesday is unveiling integrated Fibre Channel and SAN (storage area network) connectivity via internal switches to its line of BladeCenter ultradense servers.

Currently, the BladeCenter and BladeCenter HS20 offer two external 80GB SCSI drives or access to network-attached storage, as well as dual GB Ethernet connectivity.

Now, 2GB connections are embedded into every blade, with the connection to a SAN coming via a single cable from dual GB switches in the back of the chassis, Thomas Dougherty, director of IBM blade server strategy, said in an interview in the companys New York offices.

The upgrade storage capabilities are part of a series of planned enhancements to an architecture that IBM introduced last year, Dougherty said. Other enhancements will include blades powered by the companys 64-bit Power 4 chip—currently the servers run on one or two 32-bit Intel Corp. Xeon processors up to 2.4GHz—and blades with four processors. Those should roll out by the end of the year, he said.

Hewlett-Packard Co., of Palo Alto, Calif., next month will release a four-processor blade, the BL40p, which also will feature SAN connectivity.

Currently, 14 BladeCenter blades fit into a 7U box. A standard 42U rack can hold 84 blades. They can run Windows and Linux, and the ability to run AIX 5L and OS/400 will also be offered in the future, he said.

Dougherty said IBM is seeing a spike in the demand for BladeCenter. The company recently shipped its 5,000 BladeCenter, less than three months after it was introduced to the market.

Server consolidation and high availability are key factors playing into the demand, he said.

"They not only want to save money, but they also want to simplify their infrastructure," Dougherty said.

Ultradense blades take up less room than traditional servers, take away any single point of failure—if one blade fails, it can be switched out with a new one without having to bring the system down—and can be managed from a single point via management software.