Re: plans for bitmap indexes?

From: Hannu Krosing <hannu(at)tm(dot)ee>
To: Yann Michel <yann-postgresql(at)spline(dot)de>
Cc: Dave Page <dpage(at)vale-housing(dot)co(dot)uk>, Reini Urban <rurban(at)x-ray(dot)at>, pgsql-hackers(at)postgresql(dot)org
Subject: Re: plans for bitmap indexes?
Date: 2004-10-08 10:06:26
Message-ID: 1097229986.2632.9.camel@fuji.krosing.net
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On R, 2004-10-08 at 12:45, Yann Michel wrote:
> Hi,
>
> On Fri, Oct 08, 2004 at 10:09:18AM +0100, Dave Page wrote:
> > I think what Reini was asking was why do you think you need bitmap
> > indexes as opposed to any existing type?
>
> due to I'm developing a datawarehousing application we have lots of
> fact-data in our central fact-table. As I know how to improve
> performance on Oracle based datawarehouses, I'm used to add bitmap
> indexes for atributes having only a few distinct values.
> So I was looking for any comparable indexing technology but didn't find
> any so far.

There is currently no suitable index type for this type of queries (huge
tables with a few distinct values).

You may try to optimise performance by partitioning your fact tables on
these few dimension values by using table inheritance or union all
views.

There was a discussion on partitioning postgres tables on
pgsql-performance list a few days ago.

-------------
Hannu

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Oleg Bartunov 2004-10-08 10:09:54 Re: plans for bitmap indexes?
Previous Message Yann Michel 2004-10-08 09:45:28 Re: plans for bitmap indexes?