Received: from localhost (unknown [200.46.204.187]) by postgresql.org (Postfix) with ESMTP id 8FB662E3D16 for ; Fri, 25 Jan 2008 04:18:53 -0400 (AST) Received: from postgresql.org ([200.46.204.71]) by localhost (mx1.hub.org [200.46.204.187]) (amavisd-maia, port 10024) with ESMTP id 87234-09 for ; Fri, 25 Jan 2008 04:18:43 -0400 (AST) Received: from mx01.bfk.de (mx01.bfk.de [193.227.124.2]) by postgresql.org (Postfix) with ESMTP id E55952E3D20 for ; Fri, 25 Jan 2008 04:18:45 -0400 (AST) Received: from mx00.int.bfk.de ([10.119.110.2]) by mx01.bfk.de with esmtps (TLS-1.0:RSA_AES_256_CBC_SHA1:32) id 1JIJlp-0001Lo-0q; Fri, 25 Jan 2008 09:18:41 +0100 Received: from fweimer by bfk.de with local id 1JIJlp-0004jH-Ri; Fri, 25 Jan 2008 09:18:41 +0100 To: Chris Browne Cc: pgsql-performance@postgresql.org Subject: Re: Making the most of memory? References: <4796BE84.9030803@centriccrm.com> <47979C15.3040406@burntmail.com> <4797A8B3.5040605@janestcapital.com> <82tzl39u55.fsf@mid.bfk.de> <60r6g7kxax.fsf@dba2.int.libertyrms.com> From: Florian Weimer Date: Fri, 25 Jan 2008 09:18:41 +0100 In-Reply-To: <60r6g7kxax.fsf@dba2.int.libertyrms.com> (Chris Browne's message of "Thu, 24 Jan 2008 11:23:02 -0500") Message-ID: <824pd28gim.fsf@mid.bfk.de> MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable X-Virus-Scanned: Maia Mailguard 1.0.1 X-Archive-Number: 200801/230 X-Sequence-Number: 28469 * Chris Browne: >> A dedicated RAID controller with battery-backed cache of ssuficient >> size and two mirrored disks should not perform that bad, and has the >> advantage of easy availability. > > That won't provide as "souped up" performance as "WAL on SSD," and > it's from technical people wishing for things that some of those > things actually emerge... For WAL (I/O which is mostly sequential), the proposed approach isn't that bad. You can easily get more than 15,000 write transactions per second, with a single thread. Good luck finding a SSD NAS with a 60 usec round-trip time. 8-> Something which directly speaks SATA or PCI might offer comparable performance, but SSD alone isn't sufficient. --=20 Florian Weimer BFK edv-consulting GmbH http://www.bfk.de/ Kriegsstra=DFe 100 tel: +49-721-96201-1 D-76133 Karlsruhe fax: +49-721-96201-99