The new blade server will let Sun re-enter a market that's currently dominated by Dell, Egenera, Hewlett-Packard, and IBM.
A planned expansion of Sun Microsystems' Galaxy platform of x86-based servers scheduled for midyear will include a blade option, returning the company to a fast-growing market it exited two years ago.
The new line of Galaxy servers, first introduced by Sun in September, will expand the platform to include an eight-socket version as well as a new blade server, says John Fowler, executive VP of Sun's Network Systems Group.
The new Galaxy blade server will allow Sun to re-enter the blade server market that's currently dominated by Dell, Egenera, Hewlett-Packard, and IBM. Sun introduced its first x86-based blade server in 2002 with a design that used an Intel Xeon processor. It later withdrew support for the platform. Since then the blade server market has begun to grow more rapidly.
The Galaxy server platform was created by Sun co-founder Andy Bechtolsheim, who returned to Sun in 2004 to head up the Galaxy effort. Sun introduced three versions of the server in September in one- and two-socket rack deployments.
The new eight-socket Galaxy systems are already in use in several test sites, including at the Tokyo Institute of Technology, which has created a supercomputing grid of more than 5,000 dual-core Opteron processors from Advanced Micro Devices Inc. based on a network of the eight-socket Galaxy servers. By using eight dual-core Opteron processors in each server, Sun is able to effectively create a 16-way server platform.
The general availability of the eight-socket Galaxy servers will let Sun more effectively address the server consolidation and database server market, Fowler says.
Sun's re-entry into the blade server market is well-timed, Fowler says, as the market has matured significantly since Sun "retired" the company's previous blade offering. But blades remain less than 3% of the total server market.
He says the Galaxy blade will provide "differentiated value" over other blade servers currently in the market. Fowler says existing blade server offerings have failed to fully address the real reasons customers are beginning to implement blades--namely, to get greater server density while reducing management complexity.
"Over time blade servers are going to be a substantial percentage of the computing market, and that is without a doubt," Fowler says. "The blade products are going to need to continue to improve before they can displace rackmount servers as the default building block … and that is what we have gone after with this design."