I've been tracking the performance of our DB query statements
across a number of fairly high-volume pg clusters for several
years (combined 2700 tps, ~1.3TB). Last year, we started
migrating off HP-UX IA64 servers running pg 8.1.x onto Linux
quadcore x86_64 Blade servers running pg 8.3.x while running on
a high-grade SAN. Our average, overall query performance has
improved by a very pleasant ~75%. But I'm curious why 'commit'
statements (as well as certain update statements) seem to have
actually degraded (1ms vs 5ms on avg, 2ms vs 14ms in the 95th
percentile, etc). Any ideas?
TIA.
Ed