Re: Tablespaces - Mailing list pgsql-hackers

From Greg Stark
Subject Re: Tablespaces
Date
Msg-id 87y8mwvizl.fsf@stark.xeocode.com
Whole thread Raw
In response to Re: Tablespaces  (Dennis Bjorklund <db@zigo.dhs.org>)
List pgsql-hackers
Dennis Bjorklund <db@zigo.dhs.org> writes:

> On Thu, 26 Feb 2004, Gavin Sherry wrote:
> 
> > Comments? Questions? Suggestions?
> 
> Is that plan that in the future one can split a single table into 
> different table spaces? Like storing all rows with year < 1999 in one 
> tablespace and the rest in another?

That's a separate orthogonal feature called "partitioned tables". There's some
amount of resistance to the idea amongst postgres people, and there's nobody
who has spoken up interested in implementing it, but there's also lots of
interest from users. A good patch would probably go a long way to convincing
people :)

Table spaces are being able to store different tables in different physical
locations on disk. A first version of this has actually been implemented for
7.5 using symlinks.

Partitioned tables and tablespaces do indeed have a certain amount of synergy.
But even in a single tablespace your example makes sense.

> With the rule system and two underlying tables one could make it work by 
> hand I think.

The rule system could be used to do this, but there was some discussion of
using inherited tables to handle it. However neither handles the really hard
part of detecting queries that use only a part of the table and taking that
into account in generating the plan.

-- 
greg



pgsql-hackers by date:

Previous
From: Shridhar Daithankar
Date:
Subject: Re: Improving postgresql.conf
Next
From: Tom Lane
Date:
Subject: Re: simple_heap_update: tuple concurrently updated -- during INSERT