Re: Compression of full-page-writes

On Tue, Dec 30, 2014 at 01:27:44PM +0100, Andres Freund wrote:
> On 2014-12-30 21:23:38 +0900, Michael Paquier wrote:
> > On Tue, Dec 30, 2014 at 6:21 PM, Jeff Davis <pgsql(at)j-davis(dot)com> wrote:
> > > On Fri, 2013-08-30 at 09:57 +0300, Heikki Linnakangas wrote:
> > >> Speeding up the CRC calculation obviously won't help with the WAL volume
> > >> per se, ie. you still generate the same amount of WAL that needs to be
> > >> shipped in replication. But then again, if all you want to do is to
> > >> reduce the volume, you could just compress the whole WAL stream.
> > >
> > > Was this point addressed?
> > Compressing the whole record is interesting for multi-insert records,
> > but as we need to keep the compressed data in a pre-allocated buffer
> > until WAL is written, we can only compress things within a given size
> > range. The point is, even if we define a lower bound, compression is
> > going to perform badly with an application that generates for example
> > many small records that are just higher than the lower bound...
> > Unsurprisingly for small records this was bad:
>
> So why are you bringing it up? That's not an argument for anything,
> except not doing it in such a simplistic way.

I still don't understand the value of adding WAL compression, given the
high CPU usage and minimal performance improvement. The only big
advantage is WAL storage, but again, why not just compress the WAL file
when archiving.

I thought we used to see huge performance benefits from WAL compression,
but not any more? Has the UPDATE WAL compression removed that benefit?
Am I missing something?

--
Bruce Momjian <bruce(at)momjian(dot)us> http://momjian.us
EnterpriseDB http://enterprisedb.com

+ Everyone has their own god. +