Re: Slow count(*) again...

Greg Smith <greg@2ndquadrant.com>

From: Greg Smith <greg@2ndquadrant.com>

To: Samuel Gendler <sgendler@ideasculptor.com>

Cc: Joshua Tolley <eggyknap@gmail.com>, Craig Ringer <craig@postnewspapers.com.au>, mladen.gogala@vmsinfo.com, Neil Whelchel <neil.whelchel@gmail.com>, "pgsql-performance@postgresql.org" <pgsql-performance@postgresql.org>

Date: 2010-10-12T06:39:32Z

Lists: pgsql-hackers, pgsql-performance

Samuel Gendler wrote:
> I spent some time going through the various tuning docs on the wiki 
> whie bringing some new hardware up and I can't remember seeing any 
> discussion of tweaking read-ahead at all in the normal 
> performance-tuning references.  Do you have any documentation of the 
> kinds of tweaking you have done and its effects on different types of 
> workloads?

Much of my recent research has gone into the book you'll see plugged 
below rather than the wiki.  The basics of read-ahead tuning is that you 
can see it increase bonnie++ sequential read results when you increase 
it, to a point.  Get to that point and stop and you should be in good shape.

-- 
Greg Smith, 2ndQuadrant US greg@2ndQuadrant.com Baltimore, MD
PostgreSQL Training, Services and Support  www.2ndQuadrant.us
Author, "PostgreSQL 9.0 High Performance"    Pre-ordering at:
https://www.packtpub.com/postgresql-9-0-high-performance/book