|
From: Ann W. H. <aha...@ib...> - 2007-01-06 17:30:19
|
= m. Th = wrote: > > http://www.postgresql.org/docs/8.1/static/catalog-pg-autovacuum.html Autovacuum is irrelevant as we already have on-line garbage collection either cooperative or through a secondary thread. Yes, their vacuum does update statistics and Postgres keeps much more detailed index statistics (e.g. histograms) than does Firebird. Postgres also delays optimization until query execution - and revises it during execution to take advantage of values produces by an outer loop for example. > > Something related is here > http://www.postgresql.org/community/survey.41 This one is interesting.... We already have five and a half of the eight most wanted new features in Postgres ... the world's most advanced open source database. Here's the list. Allowing index scans to use an intermediate in-memory bitmap Add two-phase commit Create a new role system that replaces users and groups Automatically use indexes for MIN() and MAX() Move /contrib/pg_autovacuum into the main server Add shared row level locks using SELECT ... FOR SHARE Add dependencies on shared objects, specifically roles Add an option to psql to automagically rollback errors without aborting the transaction > > http://archives.postgresql.org/pgsql-bugs/2005-07/msg00101.php which explains that autovacuum isn't sensitive to the amount of data change, but considers one update equivalent to another in reaching the threshold for garbage collection. Firebird doesn't have that problem at all. > > > > Perhaps helps also > > http://www.iiug.org/software/archive/utils2_ak > Doesn't seem relevant to me. Regards, Ann |