IBM tops server speed test

New top-end Power5-based Unix server has commanding lead in performance test of servers running a large database.

Stephen Shankland principal writer
Stephen Shankland has been a reporter at CNET since 1998 and writes about processors, digital photography, AI, quantum computing, computer science, materials science, supercomputers, drones, browsers, 3D printing, USB, and new computing technology in general. He has a soft spot in his heart for standards groups and I/O interfaces. His first big scoop was about radioactive cat poop.
Expertise processors, semiconductors, web browsers, quantum computing, supercomputers, AI, 3D printing, drones, computer science, physics, programming, materials science, USB, UWB, Android, digital photography, science Credentials
  • I've been covering the technology industry for 24 years and was a science writer for five years before that. I've got deep expertise in microprocessors, digital photography, computer hardware and software, internet standards, web technology, and other dee
Stephen Shankland
IBM's new top-end Power5-based Unix server has a commanding lead in a performance test of servers running a large database.

A $16.7 million configuration of the company's eServer p5-595 performed 3.21 million transactions per minute on the TPC-C test from the Transaction Processing Performance Council, IBM said Wednesday. The server will become widely available Thursday.

The score topped two previous records: the fastest cluster, held by Hewlett-Packard Itanium machines running Red Hat Linux at 1.18 million transactions per minute, and the fastest single system, an IBM Power4 server running the AIX operating system at 1.03 million transactions per minute.

IBM's system had 32 1.9GHz Power5 processors, each with dual-processing engines called cores; 2 terabytes of memory; IBM's DB2 database software; and 6,548 hard drives in 45 DS4500 storage systems and 490 accompanying DS4000 storage expansion systems.

The TPC-C is no longer a good reflection of real-world computing loads--its configurations use absurdly large amounts of storage equipment, for example--but it's still the best server test around, said Illuminata analyst Jonathan Eunice. One alternative, the sales and distribution test running SAP's business software, is useful, but it explicitly prohibits the inclusion of pricing information, which is a basic part of TPC-C.

"It's easy to say TPC-C is meaningless, but what's the alternative? The problem is, if you don't do TPC-C, you're left with nada," Eunice said.