Scott Shattuck <ss@technicalpursuit.com> writes:
> Robert Treat wrote:
>> I don't think this is entirely true. On tables that have large numbers
>> of inserts, but no updates or deletes, you do not need to run vacuum.
> In my experience I've seen tables with numerous indexes continue to
> benefit greatly from vacuum/vacuum full operations when large volumes of
> inserts are performed. This is true even when the update/delete activity
> on the base table itself is manageable.
This is hard to believe, as VACUUM does not even touch the indexes
unless it has found deletable tuples --- and I am quite certain that
btree indexes, at least, do not do any VACUUM-time reorganization beyond
deleting deletable entries. (I wouldn't swear to it one way or the
other for GiST though.) Robert's opinion coincides with what I know of
the code.
regards, tom lane