No UNIX Taxation Without Performance Justification!

Dump the UNIX tax? That's what Unisys recommended when the company announced its newest SQL Server benchmark results on June 12. SQL Server crossed an important threshold on the SAP Sales and Distribution (SD) Standard Application Benchmark when Unisys demonstrated that a 32-CPU Unisys ES7000 running SQL Server 2000 and Windows 2000 could outperform a 64-CPU Sun E10000 running Oracle (see the Unisys Web site for the press release).

"The message emerging from this achievement is no UNIX taxation without performance justification," said Mark Feverston, vice president of Unisys Server Programs. "With the Unisys ES7000 outperforming even a 64 x 64-bit UNIX server, enterprises now stand to benefit from a new brand of large-scale server economics."

In the March 8 issue of SQL Server Magazine UPDATE, I discussed a 16-CPU score that Unisys posted for this same benchmark. (See the SQL Server Magazine Web site for the full article.) At that time, I said that I had been anticipating 16- and 32- way benchmark scores for months—since last summer's release of SQL Server 2000. "Bit by bit, SQL Server systems are closing the gap with the world's fastest 64-CPU UNIX database servers," I wrote then. "Unquestionably, 32-CPU SQL Server 2000 benchmarks running on 'Wintel mainframes' will be an important milestone in the quest for single-server performance parity. Will the gap close completely when someone publishes 32-CPU scores sometime in the near future? Perhaps not, but the gap is closing awfully fast."

The gap is now closed. No UNIX taxation without performance justification? It might not be time to dump your UNIX databases in Boston Harbor, but the performance landscape has changed forever now that a 32-CPU Wintel box is performance-competitive with a 64-CPU UNIX server—for a fraction of the price. Some readers question my coverage of high-end benchmarks, pointing out that super high-end tests have little relevance to the systems their companies run. I understand their point. Most people don't run the world's most intensive transaction-processing systems. But the super high-end database benchmarks are more relevant than you might realize—even for typical database teams that work for typical companies. I'll try to convince you of that fact in next week's column.

On a different note, Microsoft TechEd 2001 wrapped up in Atlanta as I wrote this week's column. I'll share my thoughts about TechEd in an upcoming article, but I'd also like to hear what you thought about the conference. Send me the three most important SQL Server-related "things" you saw, learned, or heard at TechEd. I'll compile your answers and share the best in an upcoming issue of SQL Server Magazine UPDATE.

Dump the UNIX tax? That's what Unisys recommended when the company announced its newest SQL Server benchmark results on June 12. SQL Server crossed an important threshold on the SAP Sales and Distribution (SD) Standard Application Benchmark when Unisys demonstrated that a 32-CPU Unisys ES7000 running SQL Server 2000 and Windows 2000 could outperform a 64-CPU Sun E10000 running Oracle (see the Unisys Web site for the press release).

"The message emerging from this achievement is no UNIX taxation without performance justification," said Mark Feverston, vice president of Unisys Server Programs. "With the Unisys ES7000 outperforming even a 64 x 64-bit UNIX server, enterprises now stand to benefit from a new brand of large-scale server economics."

In the March 8 issue of SQL Server Magazine UPDATE, I discussed a 16-CPU score that Unisys posted for this same benchmark. (See the SQL Server Magazine Web site for the full article.) At that time, I said that I had been anticipating 16- and 32- way benchmark scores for months—since last summer's release of SQL Server 2000. "Bit by bit, SQL Server systems are closing the gap with the world's fastest 64-CPU UNIX database servers," I wrote then. "Unquestionably, 32-CPU SQL Server 2000 benchmarks running on 'Wintel mainframes' will be an important milestone in the quest for single-server performance parity. Will the gap close completely when someone publishes 32-CPU scores sometime in the near future? Perhaps not, but the gap is closing awfully fast."

The gap is now closed. No UNIX taxation without performance justification? It might not be time to dump your UNIX databases in Boston Harbor, but the performance landscape has changed forever now that a 32-CPU Wintel box is performance-competitive with a 64-CPU UNIX server—for a fraction of the price. Some readers question my coverage of high-end benchmarks, pointing out that super high-end tests have little relevance to the systems their companies run. I understand their point. Most people don't run the world's most intensive transaction-processing systems. But the super high-end database benchmarks are more relevant than you might realize—even for typical database teams that work for typical companies. I'll try to convince you of that fact in next week's column.

On a different note, Microsoft TechEd 2001 wrapped up in Atlanta as I wrote this week's column. I'll share my thoughts about TechEd in an upcoming article, but I'd also like to hear what you thought about the conference. Send me the three most important SQL Server-related "things" you saw, learned, or heard at TechEd. I'll compile your answers and share the best in an upcoming issue of SQL Server Magazine UPDATE.