Received: from localhost (maia-2.hub.org [200.46.204.187]) by postgresql.org (Postfix) with ESMTP id 3C8219FB2E8 for ; Sat, 31 Mar 2007 12:54:22 -0300 (ADT) Received: from postgresql.org ([200.46.204.71]) by localhost (mx1.hub.org [200.46.204.187]) (amavisd-maia, port 10024) with ESMTP id 72014-01 for ; Sat, 31 Mar 2007 12:54:18 -0300 (ADT) X-Greylist: from auto-whitelisted by SQLgrey-1.7.4 Received: from cse.ohio-state.edu (mail.cse.ohio-state.edu [164.107.123.5]) by postgresql.org (Postfix) with ESMTP id 4FB629FB2B4 for ; Sat, 31 Mar 2007 12:54:19 -0300 (ADT) Received: from [164.107.116.241] (jason.cse.ohio-state.edu [164.107.116.241]) (authenticated bits=0) by cse.ohio-state.edu (8.12.11.20060308/8.12.11) with ESMTP id l2VFsIRu022554; Sat, 31 Mar 2007 11:54:18 -0400 (EDT) Message-ID: <460E8427.5090502@cse.ohio-state.edu> Date: Sat, 31 Mar 2007 11:54:15 -0400 From: Xiaoning Ding User-Agent: Thunderbird 1.5.0.10 (X11/20070301) MIME-Version: 1.0 To: Christopher Browne CC: pgsql-performance@postgresql.org Subject: Re: scalablility problem References: <460D7220.9010302@cse.ohio-state.edu> <874po2tdlx.fsf@wolfe.cbbrowne.com> In-Reply-To: <874po2tdlx.fsf@wolfe.cbbrowne.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: Maia Mailguard 1.0.1 X-Archive-Number: 200703/467 X-Sequence-Number: 23872 Christopher Browne wrote: > Quoth dingxn@cse.ohio-state.edu (Xiaoning Ding): >> When I run multiple TPC-H queries (DBT3) on postgresql, I found the system >> is not scalable. My machine has 8GB memory, and 4 Xeon Dual Core processor >> ( 8 cores in total). OS kernel is linux 2.6.9. Postgresql is 7.3.18. I > >> I think it might be caused by some contentions. But I do not know postgresql >> much. May anybody give me some clue to find the reasons? > > Two primary issues: > > 1. You're running a horrendously ancient version of PostgreSQL. The > 7.3 series is Really Old. Newer versions have *enormous* > improvements that are likely to be *enormously* relevant. > > Upgrade to 8.2. 8.2 is really much better. > > 2. There are known issues with the combination of Xeon processors and > PAE memory addressing; that sort of hardware tends to be *way* less > speedy than the specs would suggest. I think PAE slows each query process. It would not affect scalability. > There have been "context switching" issues on this sort of hardware > that are enormously worsened if you're running on a version of > PostgreSQL that is 4 major releases out of date. How does PG 8.2 address this issue? by setting processor affinity? Thanks! Xiaoning