Re: Proposal of PITR performance improvement for 8.4.

On Thu, 2008-10-30 at 09:46 +0900, Koichi Suzuki wrote:

> I'm not sure if blocks reffered from all WAL records in single WAL
> segment can fit kernel cache. This is why current pg_readahead
> returns the last LSN and require starting LSN. So far, with FPW, it
> seems that we can prefetch all the pages in a WAL segment. So it
> will be okay for archive log compressed with pg_compresslog. I'll
> test if it's okay in the case full_page_writes=off.

I'd prefer to be able to specify max_readahead_pages than have to
control things at a micro level like that. If you have lots of memory
you can set that higher.

> Anyway, I'd like to keep my proposal for 8.4 and continue the test and
> evaluation to report to the mailing list.
>
> I'll also change the whole code to run in the core.

OK, I quite liked the idea of a separate program. That allows it to work
with 8.3 as well as 8.4. No problem with it being in core at all.

As ever, good thinking, good patch.

--
Simon Riggs www.2ndQuadrant.com
PostgreSQL Training, Services and Support