Date: Sun, 22 Jan 2006 11:49:36 +0100 (CET) From: Oliver Fromme <olli@lurza.secnetix.de> To: freebsd-database@FreeBSD.ORG Subject: Re: Horrible PostgreSQL performance with NFS Message-ID: <200601221049.k0MAnaVh089477@lurza.secnetix.de> In-Reply-To: <20060119011550.GN17896@decibel.org>
next in thread | previous in thread | raw e-mail | index | archive | help
Jim C. Nasby wrote: > On Wed, Jan 18, 2006 at 05:11:15PM -0800, Brooks Davis wrote: > > On Wed, Jan 18, 2006 at 06:58:47PM -0600, Jim C. Nasby wrote: > > > On Sat, Jan 14, 2006 at 06:42:02AM -0800, Arne Woerner wrote: > > > > Did you do those "dd" tests with small block sizes (like 1byte: > > > > bs=1), like somebody on one of those lists suggests, too? Then we > > > > could see, if there is a high latency that ruins everything... > > > > > > FYI, PostgreSQL does 8kB I/O by default. This can only be changed by > > > modifying a header file. > > > > That's definitely small in my book (certainly compared to the 1MB block > > size the first responder suggested), knowing that, you should > > definitely do dd tests with 8k blocks since that's the best performance you > > are likely to get. > > Agreed, it is small from a OS/filesystem viewpoint, but it's also > nowhere near 1 byte which is the test that had been suggested. :) FWIW, there is _no_ blocksize that can be used wit dd(1) that benchmarks disks the same way a database does. Unless our dd(1) implemetation grew a random-seek option that I missed. :-) But seriously... dd(1) is NOT a benchmark. Never ever. Especially if you're interested in database performance. Best regards Oliver -- Oliver Fromme, secnetix GmbH & Co. KG, Marktplatz 29, 85567 Grafing Dienstleistungen mit Schwerpunkt FreeBSD: http://www.secnetix.de/bsd Any opinions expressed in this message may be personal to the author and may not necessarily reflect the opinions of secnetix in any way. "If you aim the gun at your foot and pull the trigger, it's UNIX's job to ensure reliable delivery of the bullet to where you aimed the gun (in this case, Mr. Foot)." -- Terry Lambert, FreeBSD-hackers mailing list.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200601221049.k0MAnaVh089477>