On Fri, 2006-02-17 at 11:06, Emi Lu wrote:
> >>>>In another way, whenever we "delete/truncate and then insert" data into
> >>>>a table, it is better to "vacuum anaylze"?
> >>>>
> >>>>
> >>>>
> >>>>
> >>>You shouldn't need a VACUUM if you haven't yet done any updates or
> >>>deletes since the TRUNCATE. An ANALYZE seems like a good idea, though.
> >>>(You could get away without ANALYZE if the new data has essentially the
> >>>same statistics as the old, but if you're making only minor changes, why
> >>>are you using this technique at all ...)
> >>>
> >>>
> >>>
> >>>
> >>After truncate table A, around 60,000 will be inserted. Then a
> >>comparision will be done between table A and table B. After that, table
> >>B will be updated according to the comparision result. Records inserted
> >>into table A is increasing everyday.
> >>
> >>So, your suggestion is that after the population of table A, the query
> >>planner should be able to find the most efficient query plan because we
> >>do truncate but not delete, and we do not need to do vacuum analyze at
> >>all, right?
> >>
> >>
> >
> >no. the suggestion was that a VACUUM is not needed, but
> >that an ANALYZE might be.
> >
> >
>
> Thank you gnari for your answer. But I am a bit confused about not running vacuum but only "analyze". Can I seperate
thesetwo operations? I guess "vacuum analyze" do both vacuum and analyze. Or "EXPLAIN ANALYZE" can do it for me?
Yeah, vacuum analyze is kind of a leftover from the olden days when you
could only run an analyze as part of a vacuum command. analyze has been
it's own command for quite some time now.
