X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 8092B52875 for ; Thu, 2 Jun 2005 11:47:23 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 22057-08 for ; Thu, 2 Jun 2005 14:47:19 +0000 (GMT) Received: from web51004.mail.yahoo.com (web51004.mail.yahoo.com [206.190.38.135]) by svr1.postgresql.org (Postfix) with SMTP id E88D452862 for ; Thu, 2 Jun 2005 11:47:17 -0300 (ADT) Received: (qmail 15004 invoked by uid 60001); 2 Jun 2005 14:47:21 -0000 DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=s1024; d=yahoo.com; h=Message-ID:Received:Date:From:Subject:To:Cc:In-Reply-To:MIME-Version:Content-Type:Content-Transfer-Encoding; b=MZzPM2GEqHh0Q0fqeQJf+dPkEJpIsYbQUvPSojdQwHv+5lgzLNF8GmUmFYLGnsuqMXdgSbJNr7eMlb0ESHFwJuw1szZjVrKuS//6EkjyoJrGQOZp3ertjnTeHdBvezfjpWzg4bgO6sp83RH6vgb9EeKhpDnvVf+M+1PivTNmJgU= ; Message-ID: <20050602144721.15002.qmail@web51004.mail.yahoo.com> Received: from [203.101.103.131] by web51004.mail.yahoo.com via HTTP; Thu, 02 Jun 2005 07:47:21 PDT Date: Thu, 2 Jun 2005 07:47:21 -0700 (PDT) From: Himanshu Baweja Subject: Re: Moving pg_xlog To: Tom Lane Cc: pgsql-performance@postgresql.org In-Reply-To: <13548.1117722203@sss.pgh.pa.us> MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="0-162673045-1117723641=:14988" Content-Transfer-Encoding: 8bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.522 tagged_above=0 required=5 tests=AWL, DNS_FROM_RFC_ABUSE, DNS_FROM_RFC_POST, HTML_20_30, HTML_MESSAGE X-Spam-Level: * X-Archive-Number: 200506/41 X-Sequence-Number: 12679 --0-162673045-1117723641=:14988 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit Tom Lane wrote: >It seems highly unlikely that putting more stuff on the xlog disk will >improve performance --- at least not if your bottleneck is update speed. Tom you are right.. i did some testing... 1) default config--- xlog on disk1 and data on disk2=> 27 mins and 22 secs 2) xlog and some tables on disk1 and rest of tables on disk2=> 28 mins and 38 secs but the most startling of the results is.... 3) xlog on disk1 and half the tables on partition 1 of disk2 and other half on partition 2 of disk2 24 mins and 14 secs ?????????? shouldnt moving data to diff partitions degrade performance instead of enhancing it.... also in configuration 1, my heap_blks_hit/heap_blks_fetched was good enough.... but in configuration 3, its was really low.. something of the order of 1/15... still the performance improved.... any ideas..... does moving across partitions help... Regards Himanshu __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com --0-162673045-1117723641=:14988 Content-Type: text/html; charset=iso-8859-1 Content-Transfer-Encoding: 8bit
Tom Lane <tgl@sss.pgh.pa.us> wrote:
>It seems highly unlikely that putting more stuff on the xlog disk will
>improve performance --- at least not if your bottleneck is update speed.
 
Tom you are right.. i did some testing...
1) default config--- xlog on disk1 and data on disk2=>
    27 mins and 22 secs
2) xlog and some tables on disk1 and rest of tables on disk2=>
    28 mins and 38 secs
 
but the most startling of the results is....
3) xlog on disk1 and half the tables on partition 1 of disk2 and other half on partition 2 of disk2
   24 mins and 14 secs
 
 ??????????
shouldnt moving data to diff partitions degrade performance instead of enhancing it....
 
also  in configuration 1, my heap_blks_hit/heap_blks_fetched was good enough....
but in configuration 3, its was really low.. something of the order of 1/15...
still the performance improved....
 
any ideas.....
does moving across partitions help...
 
Regards
Himanshu
 

__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com --0-162673045-1117723641=:14988--