Re: Linux Software RAID 1 Performance (was:Re: Re: Slower on Solaris)

> RAID 1 doesn't change the PostgreSQL performance on the regression tests
> significantly. This machine, on multiple runs of 'time ./pg_regress
> --schedule=parallel_schedule' on the same UDMA66 drives without RAID 1
posted
> an average 'real' number of 44 seconds. With RAID 1 (and the same
drives,
> controllers, OS, etc) posts an average of 42 seconds -- not
statistically
> significant. Yes, $PGDATA was on a RAID device... :-).

One of the reasons is that software RAID does depend on the CPU even
more than normal IDE - and when your database is being hit, the CPU is
generally getting utilized pretty decently. My personal preference is to
use hardware RAID 5 for redundancy, ensure that there's plenty of RAM to
keep all database files in cache, and turn off fsync(). Even when our
database server is maxing out all four processors, the lights on the RAID
array only blink *occasionally*. : )

steve