Re: Higher TOAST compression.

From: "Kevin Grittner" <Kevin(dot)Grittner(at)wicourts(dot)gov>
To: "Laurent Laborde" <kerdezixe(at)gmail(dot)com>
Cc: <jd(at)commandprompt(dot)com>, "PostgreSQL-development" <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: Higher TOAST compression.
Date: 2009-07-17 20:40:23
Message-ID: 4A609B670200002500028947@gw.wicourts.gov
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Laurent Laborde <kerdezixe(at)gmail(dot)com> wrote:
> Kevin Grittner<Kevin(dot)Grittner(at)wicourts(dot)gov> wrote:

>> How about two new ALTER TABLE actions:
>>
>> ALTER [ COLUMN ] column SET COMPRESSION_THRESHOLD integer
>> ALTER [ COLUMN ] column SET EXTERNAL_THRESHOLD integer

>> Laurent, would something like this address your needs?

> Certainly !
> We already alter storage type on some column, adding some more won't
> be a problem. :)
>
> But... on which version are you planning to do that ?

The patch, if there's consensus that it's a good idea, would be for
8.5. Since it is new functionality, there wouldn't be a back-port to
prior releases. Of course, I wouldn't be starting to work on such a
patch until after our current code commit phase, which ends August
15th.

> We stay on Pg 8.3 until the slony developpers find a better upgrade
> solution.
>
> The proposed solution sound really good to me.
> But, for now, if i could have a simple patch for 8.3 (eg: changing a
> #define in the source code), i'd be very happy :)
>
> Is it ok to just change TOAST_TUPLES_PER_PAGE ?

The thing that worries me about that is that it would tend to force
more data to be stored out-of-line, which might *increase* your I/O;
since the whole point of this exercise is to try to *decrease* it,
that seems pretty iffy. However, once we get to the end of code
commit, I might be able to give you a little one-off patch that would
be more aggressive about compression without affecting out-of-line
storage. Hard-coded, like what you're talking about, but with a
little more finesse.

-Kevin

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Bernd Helmle 2009-07-17 20:49:40 Re: commitfest application question
Previous Message Alan Li 2009-07-17 20:35:33 MIN/MAX optimization for partitioned table