Re: Simulate database fragmentation - Mailing list pgsql-performance

From Jim C. Nasby
Subject Re: Simulate database fragmentation
Date
Msg-id 20070527160538.GO92628@nasby.net
Whole thread Raw
In response to Simulate database fragmentation  ("Y Sidhu" <ysidhu@gmail.com>)
List pgsql-performance
On Wed, May 23, 2007 at 11:58:06AM -0700, Y Sidhu wrote:
> Is there any easy way to take a database and add/delete records to create
> fragmentation of the records and indexes. I am trying to recreate high
> vacuum times.

Update random rows, then do a vacuum. That will result in free space in
random locations. At that point you'd probably want to update some
ranges of rows, enough so that they get forced to new pages.

A better idea might be to just insert random data.
--
Jim Nasby                                      decibel@decibel.org
EnterpriseDB      http://enterprisedb.com      512.569.9461 (cell)

Attachment

pgsql-performance by date:

Previous
From: "Jim C. Nasby"
Date:
Subject: Re: Domains versus Check Constraints
Next
From: "Jim C. Nasby"
Date:
Subject: Re: Memory allocation and Vacuum abends