On Mon, Aug 12, 2013 at 04:41:48PM -0400, David F. Skoll wrote: > On Mon, 12 Aug 2013 16:05:04 -0400 > Bruce Momjian <bruce@xxxxxxxxxx> wrote: > > > On Mon, Aug 12, 2013 at 08:33:04AM -0700, Joshua D. Drake wrote: > > > > Put the pg_xlog on spindles, they are more than fast enough and > > > won't eat up the write life of your SSDs. > > > Given its small size and need for fast fsync, I have WAL on SSDs, and > > the data on magnetic disk. You are right the WAL can generate a lot > > of writes, but just use smartctl monitoring and replace the SSD when > > needed. > > :-) Nothing like consensus from Pg luminaries. :) > > Anyway, I've decided to go with spinning disks for this server upgrade. The > factors behind my decision: > > o A server with 16 10Krpm disks will be substantially better than what we > have now, and what we have now is managing to keep up with only occasional > over-busy periods. > > o My supplier cannot easily get the 320, 710 or S3700 Intel SSD disks; he > claims a shortage of Intel SSD drives. I know too little about non-Intel > SSDs to feel comfortable with them. > > o One of our two database servers is 200km away, so swapping out SSDs on > a prophylactic basis is not particularly appetizing. Good point on the travel distance. For my server, I didn't go with a RAID controller with a battery-backed unit (BBU), so SSDs for WAL made complete sense for me. FYI, your CPUs exactly matches mine: http://momjian.us/main/blogs/pgblog/2012.html#January_20_2012 -- Bruce Momjian <bruce@xxxxxxxxxx> http://momjian.us EnterpriseDB http://enterprisedb.com + It's impossible for everything to be true. + -- Sent via pgsql-admin mailing list (pgsql-admin@xxxxxxxxxxxxxx) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-admin
