Re: Autovacuum in the backend

Gavin, People,

> I'm wondering if effort is being misdirected here. I remember when Mark
> Wong at OSDL was running pg_autovacuum during a dbt run, he was seeing
> significant performance loss -- I think on the order of 30% to 40% (I will
> try and dig up a link to the results).

It wasn't quite that bad, and the automated DBT2 is deceptive; the test
doesn't run for long enough for *not* vacuuming to be a problem. For a real
test, you'd need to do a 24-hour, or 48-hour DBT2 run.

Not that I don't agree that we need a less I/O intense alternative to VACUUM,
but it seems unlikely that we could actually do this, or even agree on a
spec, before feature freeze. Wheras integrated AV is something we *could*
do, and is widely desired.

If we do integrated AV, it should only be turned on by default at a relatively
low level. And wasn't there an issue on Windows with AV not working?

--
Josh Berkus
Aglio Database Solutions
San Francisco