Sun Eclipses IBM on SPC Test

Sun Microsystems Inc. (Nasdaq: SUNW), which has been a major supporter of the Storage Performance Council (SPC)'s benchmark testing, today tried to use those tests as proof that its entry-level Fibre Channel storage system is a better value than a similar offering from IBM Corp. (NYSE: IBM).

Sun pointed to significant advantages to scores posted by the StorEdge 3510 FC array over IBM's TotalStorage FastT600 with Turbo Option, after both systems were tested using the SPC-1 benchmark. The StorEdge 3510 turned in a 17 percent edge in performance over the FastT600, with a 33 percent better price-performance, according to SPC's industry-standard tests.

The SPC tested the StorEdge 3510 Fibre Channel system with a SunFire 6800 server and found the storage system delivered 11,049.6 input/output operations per second (IOPS). The FastT600 Turbo, tested with a Dell 6650 server and with mirrored write-cache enabled, delivered 9,099.86 IOPS.

The StorEdge 3510 registered a price/performance result of $7.95 per IOPS compared with the FastT600's price-performance score of $11.86 per IOPS. The price of Sun's unit as tested is $87,889, compared with $107,900 for the IBM system.

"We think that this is a compelling price/performance story," says Randall Sagrillo, Sun's manager of technical storage marketing.Ron Johnson, senior partner at the Evaluator Group, says the reason Sun's price/performance for the 3510 looks better is because it's underpricing the competition. "Typically, entry-level external arrays all give more than adequate performance," he says. "The significant thing is Sun has priced it very aggressively."

According to the SPC results, IBM had a slight edge in average response time. The FastT600 had an average response time of 2.3 milliseconds at 10 percent workload level and 28.46ms at 100 percent workload, whereas the StorEdge 3510's average response time was 2.97ms at 10 percent and 29.36ms at 100 percent.

What does IBM say about the tests? Not much. The company sent Byte and Switch this statement: "There are a number of variables that affect these types of benchmarks. As IBM is in the process of analyzing the configuration, pricing, and metrics used, [we] are unable to comment specifically on this test at the present time."

Sun announced the StorEdge 3510 in February 2003, two months before IBM countered with the FastT600 (see Sun Thinks Small on Storage and HP, IBM Muscle Up Midrange).

To ensure authenticity, accuracy, and compliance, the SPC requires completion of a results validation process -- including audit certification and peer review -- before benchmark results become official. IBM's system was tested in August 2003 and the results were accepted in October. Sun's results were submitted for review Monday, and have yet to be accepted.Sagrillo says potential Sun customers often ask for SPC-1 numbers when making buying decisions. "I know there is a segment of the market that cares about this, and we take it seriously here," he says.

Dave Raffo, Senior Editor, Byte and Switch