On Wed, Jun 15, 2005 at 09:07:30PM -0700, Josh Berkus wrote: > Josh, > > > Just my own two cents. First I am not knocking the work that has been on > > autovacuum. I am sure that it was a leap on its own to get it to work. > > However I will say that I just don't see the reason for it. > > I've personally seen at least a dozen user requests for "autovacuum in the > backend", and had this conversation about 1,100 times: > > NB: "After a week, my database got really slow." > Me: "How often are you running VACUUM ANALYZE?" > NB: "Running what?"
Yes, me too. I always understood autovacuum to be a way to avoid having newbies get burned by not vacuuming, and for simplifying the maintenance of lower traffic databases. I don't see people with high-traffic databases (relative to the hardware they're running on) ever using autovacuum with the current state of vacuum and autovacuum. If improvements to vacuum (unrelated to autovacuum) reduce the IO load that would be a great thing, especially for those of us dealing with 24x7 databases. (I really like the dirty bitmap suggestion - it sounds a clean way to reduce the amount of work needed). If autovacuum were extended to allow more flexible scheduling (or even to be aware of the other IO going on) then it would be of wider use - but I think the real value of autovacuum is to make sure that new users (Windows...) don't have a bad experience when they first try PG. Cheers, Steve ---------------------------(end of broadcast)--------------------------- TIP 4: Don't 'kill -9' the postmaster