Re: Partitioned tables as a poor mans columnar index?

From: Peter Hunsberger <peter(dot)hunsberger(at)gmail(dot)com>
To: marcin mank <marcin(dot)mank(at)gmail(dot)com>
Cc: pgsql-general(at)postgresql(dot)org
Subject: Re: Partitioned tables as a poor mans columnar index?
Date: 2009-10-16 21:01:01
Message-ID: cc159a4a0910161401g1f84011nc46a58bd6dcab43c@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-general

On Fri, Oct 16, 2009 at 3:31 PM, marcin mank <marcin(dot)mank(at)gmail(dot)com> wrote:
> On Fri, Oct 16, 2009 at 9:19 PM, Peter Hunsberger
> <peter(dot)hunsberger(at)gmail(dot)com> wrote:
>
>> The basic problem I have is that I have some tables that are
>> potentially very long (100,000's to millions of rows) and very skinny,
>
>> and I end up with maybe a total of 12 bits of data in each row.
>
> Are You aware that there are some 20-ish bytes of metadata for each
> row? saving 4 bytes buys You nothing. Give it up.

No, the metadata is a whole 'nother problem. I'm just talking about
fk relationships here. This isn't an isolated issue within this
particular domain. If you where to use a conventional table design,
then once the rest of the associated tables get built along with their
associated indexes you'd be looking at in the order of a terabyte for
this half of the DB...

--
Peter Hunsberger

In response to

Browse pgsql-general by date

  From Date Subject
Next Message Scott Marlowe 2009-10-16 21:51:34 Re: db not dumping properly, or at least not restoring
Previous Message Jeff Davis 2009-10-16 20:55:06 Re: Urgent Help required