In-Depth

News In Brief

IBM Linux clusters; Fujitsu servers; Windows Server 2003 performance

• By Stephen Swoyer
• 05/22/2003

IBM Corp. yesterday announced a new pre-packaged Linux-based server blade cluster, the eServer Cluster 1350. The new offering bundles server blade hardware with management software and storage.

The Cluster 1350 can be built with combinations of IBM’s BladeCenter systems, including the xSeries x335 and x345 server blades.

Big Blue provides Cluster Management Software (CSM) to address the concerns of cluster customers who need to automate repetitive tasks and error detection. CSM for the Cluster 1350 is the Linux version of software that powered Deep Blue, a chess-playing supercomputer designed by IBM.

The new Cluster 1350 supports Big Blue’s FAStT200 and FAStT700 storage systems. Like IBM’s other blade server offerings, the Cluster 1350 features an integrated fibre channel switch to facilitate integration in storage area networks.

IBM says that the new clusters will be pre-tested at the factory. They’re targeted for technical computing firms in specific industries, such as aerospace, according to Big Blue.

Fujitsu Refreshes Low-End Server Line

Fujitsu Technology Solutions yesterday refreshed the low end of its line of SPARC-powered PrimePower servers with new models.

As a developer of SPARC-based systems that run the Solaris operating environment, Fujitsu Technology Solutions Inc. has long labored in the shadow of Sun Microsystems Inc. This in spite of the fact that Fujitsu markets an implementation of SPARC—called SPARC64—that is arguably superior to Sun’s UltraSPARC III chip.

Sun originally invented SPARC (Scalable Processor ARChitecture) and handed the specification over to a standards body in 1989. Today, SPARC defines an open set of technical specifications a vendor or organization can license to produce its own SPARC microprocessors.

The new PrimePower 250 and 450 machines are available in two- and four-way configurations, and are outfitted with Fujitsu’s 1.1 GHz SPARC64 V chips. The new machines will replace Fujitsu's existing PrimePower 200 and 400 workgroup servers, which are normally populated with 600 MHz and 700 MHz SPARC64 chips.

Like IBM Corp., Fujitsu has been trickling down technologies from its high-end machines, which scale to as many as 128 SPARC64 V processors. To that end, the new workgroup servers boast support for Fujitsu’s Hardware Instruction Retry technology, which enables them to retry instructions that fail to properly execute without simply triggering an error.

Most RISC-based systems support instruction retry at the operating system level. Sun’s Solaris operating system, for example, will retry an instruction up to 17 times. Fujitsu’s Hardware Instruction Retry, however, will retry an instruction 15 times in hardware before passing it over to the operating system (in this case, Sun’s Solaris 8 or Solaris 9 operating environments, which Fujitsu ships with each of its PrimePower systems).

“The difference between a hardware retry and a software retry is that [a hardware retry] is much faster,” explained Tom Donnelly, senior product manager for the PrimePower server line with Fujitsu.

TPC-C Benchmark Heats Up, Windows Back on Topby Scott Bekker(Courtesy of ENTmag.com)

HP hoisted its Superdome-Itanium-Windows combination to the top of the premier OLTP scalability benchmark on Tuesday, about a month after originally gaining the top spot and little more than a week after IBM displaced the combo with a system based on its own AIX/RISC/DB2 stack.

The flurry of activity marks a renewed interest in the Transaction Processing Performance Council's TPC-C benchmark after several years with little meaningful activity at the top performance tier of the benchmark.

Read the full article here: http://www.entmag.com/news/article.asp?EditorialsID=5816