X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [64.117.224.130]) by svr1.postgresql.org (Postfix) with ESMTP id 60617D1B56B for ; Wed, 3 Sep 2003 14:00:01 +0000 (GMT) Received: from svr1.postgresql.org ([64.117.224.193]) by localhost (neptune.hub.org [64.117.224.130]) (amavisd-new, port 10024) with ESMTP id 98490-08 for ; Wed, 3 Sep 2003 10:59:33 -0300 (ADT) Received: from mail.iinet.net.au (mail-04.iinet.net.au [203.59.3.36]) by svr1.postgresql.org (Postfix) with SMTP id 9A4F2D1B8BA for ; Wed, 3 Sep 2003 10:59:28 -0300 (ADT) Received: (qmail 5530 invoked from network); 3 Sep 2003 13:59:26 -0000 Received: from unknown (HELO postgresql.org) (203.59.116.51) by mail.iinet.net.au with SMTP; 3 Sep 2003 13:59:26 -0000 Message-ID: <3F55F3A8.9030202@postgresql.org> Date: Wed, 03 Sep 2003 21:59:04 +0800 From: Justin Clift User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.4) Gecko/20030624 X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Cc: Jenny Zhang Subject: " ... though the pgsql version performance is not very great." Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200309/18 X-Sequence-Number: 3384 Hi everyone, Saw this earlier on today, on the mailing list of the Open Source Development Labs people who are porting their database testing suite from SAP to PostgreSQL. The comment near the end by Jenny Zhang (one of the porters), saying that "I will put a tar ball on SourceForge today, though the pgsql version performance is not very great." doesn't sound very nifty. Personally, I don't have the time to analyse why the performance isn't very good and take it forward, so I'm mentioning it here as a heads up in case someone want's something useful to sink their teeth into. :-) Regards and best wishes, Justin Clift -------- Original Message -------- Subject: Re: [osdldbt-general] DBT1 and dynamic cache Date: 02 Sep 2003 09:46:29 -0700 From: Jenny Zhang To: Wenguang.Wang@usask.ca CC: osdldbt-general@lists.sourceforge.net References: <200309021543.h82Fh0o26886@mail.osdl.org> Mark is right. DBT1 is designed to be run in several modes: dbdriver + database: each user opens a database connection, and all the transaction go to the database dbdriver + cache + database: each user opens a database connection, three transactions (bestsellers, newprodicts, and search results by subject) go to the cache, the others go to the database dbdriver + transaction manager + database: each user opens a connection to the transaction manager, which manages the transaction queue and database connection. All the transaction go to the database. dbdriver + transaction manager + database + cache: each user opens a connection to the transaction manager, which manages the transaction queue and database connection. Three transactions (bestsellers, newprodicts, and search results by subject) go to the cache, the others go to the database The pgsql version is available at bk://developer.osdl.org/dbt1. I will put a tar ball on SourceForge today, though the pgsql version performance is not very great. Jenny On Tue, 2003-09-02 at 08:42, markw@osdl.org wrote: > We do have something that simulates some of the caching effects of web > servers. It's under the 'cache' directory. If I remember correctly, it > doesn't have to be used, but we do have it working. > > Mark > > On 31 Aug, Wenguang Wang wrote: > > Hi, dbt1 designers and developers, > > > > I like the idea of eliminating the web servers in dbt1 to focus on the > > performance of DBMS. However, I have a question about whether the > > current dbt1 can really represent e-commerce workloads. > > > > TPC-W encourages the use of web caches to reduce the load on DBMS. Since > > web caches are not used in dbt1, all cachable queries have to be > > procesed by the DBMS in dbt1. This could increase the load to the > > backend DBMS by several times at least. These queries make dbt1 more > > like a TPC-H throughput test instead of an e-commerce test. Is this > > design of dbt1 intentional or is it planned to be fixed later? > > > > Thanks. > > > > > > ------------------------------------------------------- > This sf.net email is sponsored by:ThinkGeek > Welcome to geek heaven. > http://thinkgeek.com/sf > _______________________________________________ > osdldbt-general mailing list > osdldbt-general@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/osdldbt-general ------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf _______________________________________________ osdldbt-general mailing list osdldbt-general@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/osdldbt-general