graph from software raid, RAID10, no partition table, ext3, read-write load

If you haven’t heard, the Linux Plumbers Conference is happening September 17-19, 2008 in Portland, OR. It’s a gathering designed to attract Linux developers – kernel hackers, tool developers and problem solvers.

A few of us that met through the Portland PostgreSQL User Group (PDXPUG) pitched an idea for a talk on filesystem performance. We wanted to examine performance conventional wisdom and put it to the test on some sweet new hardware, recently donated for performance testing Postgres. We’re asking questions like: Is RAID5 really the worst performing configuration for a database? How much does partition alignment really matter? Is there one Linux filesystem that a DBA should always choose for best performance under any load? Is adaptive readahead all that?

Our talk was accepted, so we’ve been furiously gathering data, and drawing interesting conclusions, ever since. Gabrielle Roth and I are presenting, using the results of extensive testing conducted by Mark Wong, a database benchmarking expert and author of pg_top. We’ll be sharing 6 different assumptions about filesystem performance, tested on five different filesystems, under five types of loads generated by fio, a benchmarking tool designed by kernel hacker Jens Axboe to test I/O.