Re: Checkpoint cost, looks like it is WAL/CRC

On Fri, 2005-07-08 at 09:34 +0200, Zeugswetter Andreas DAZ SD wrote:
> >>> The point here is that fsync-off is only realistic for development
> or
> >>> playpen installations. You don't turn it off in a production
> >>> machine, and I can't see that you'd turn off the full-page-write
> >>> option either. So we have not solved anyone's performance problem.
> >
> >> Yes, this is basically another fsync-like option that isn't for
> >> production usage in most cases. Sad but true.
> >
> > Just to make my position perfectly clear: I don't want to see
> > this option shipped in 8.1.
>
> Why not ? If your filesystem buffer size matches your pg page size,
> and you have a persistent write cache, the option makes perfect sense.

I think this point needs expansion:

It is possible to run with matching page sizes, in which case the option
to not-log full page images is desirable.

The only type of crash that can cause torn pages is an OS crash, such as
a power outage. If you have an expensive server with battery backup,
persistent write cache etc then this reduces that risk considerably.

However, as long as PostgreSQL can't tell the difference between any
crash and an OS crash, we must assume the worst.

Best Regards, Simon Riggs