Re: Large Objects and and Vacuum

On 31 December 2011 00:54, Simon Windsor <simon(dot)windsor(at)cornfield(dot)me(dot)uk> wrote:
> I am struggling with the volume and number of XML files a new application is
> storing. The table pg_largeobjects is growing fast, and despite the efforts
> of vacuumlo, vacuum and auto-vacuum it keeps on growing in size

I can't help but wonder why you're using large objects for XML files?
Wouldn't a text-field be sufficient? Text-fields get toasted, that
would safe you some space.

Another option would be to use xml-fields, but that depends on whether
you have valid XML and whether you have any desire to make use of any
xml-specific features such fields provide. There will probably be a
performance hit for this.

I do realise that you can stream large objects, that's a typical
use-case for choosing for them, but with XML files that doesn't seem
particularly useful to me; after all, they're not valid if not
complete. You have to read the whole file into memory _somewhere_
before you can interpret them meaningfully. The exception to that rule
is if you're using a SAX-parser (which also explains why those parsers
usually have fairly limited features).

Of course there are valid reasons for choosing to use large objects
for XML files, I assume yours are among them. If they're not, however,
maybe you should have a thorough look at your problem again.

--
If you can't see the forest for the trees,
Cut the trees and you'll see there is no forest.