Re: Press Release

From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
To: Bruce Momjian <pgman(at)candle(dot)pha(dot)pa(dot)us>
Cc: Robert Treat <xzilla(at)users(dot)sourceforge(dot)net>, Josh Berkus <josh(at)agliodbs(dot)com>, pgsql-advocacy(at)postgresql(dot)org
Subject: Re: Press Release
Date: 2003-10-29 23:40:11
Message-ID: 23869.1067470811@sss.pgh.pa.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-advocacy

Bruce Momjian <pgman(at)candle(dot)pha(dot)pa(dot)us> writes:
> Robert Treat wrote:
>> Tom has laid out at least one case where the potential for index growth
>> exits, though I don't see it in a quick search of the archives...
>>
>> Tom, can you weigh in here?

> If you delete all but one row on every index page, I think.

It might be worth pointing out that while one entry per page would be
pretty grim, it's at least a *bounded* overhead factor. Pre-7.4 it was
possible to accumulate arbitrarily large numbers of entirely empty pages
in an index, so that the total index size could grow without bound even
when the actual number of live entries stayed about constant.

Furthermore, this worst case actually happened in a significant fraction
of real-world usages. The cases where 7.4 will degrade to a small
number of live entries per page are (probably) far more infrequent.

So I think 7.4 will greatly reduce the need for routine reindexing ...
but it's probably premature to claim that it's gone entirely.

regards, tom lane

In response to

Browse pgsql-advocacy by date

  From Date Subject
Next Message Rod Taylor 2003-10-29 23:53:10 Re: Press Release
Previous Message Josh Berkus 2003-10-29 23:39:40 Re: Press Release