On Fri, 2002-11-01 at 06:15, [EMAIL PROTECTED] wrote:
Looks like a borderline case. See the costs of the index scan and sequential scan are very similar. Since 499 covers nearly 1 in 10 tuples, it's likely found on nearly every page. This should make a sequential scan much cheaper. However, if the data is clumped together (not distributed throughout the table) than an index scan may be preferable. So... CLUSTER may be useful to you. In the future please 'explain analyze' the queries you're looking at to see actual costs as compared to the estimated cost. > 499 | 25010 > 501 | 3318 > > > before dump reload: > tradein_clients=# VACUUM VERBOSE ANALYZE email_bank_mailing_lists; > NOTICE: --Relation email_bank_mailing_lists-- > NOTICE: Pages 3583: Changed 0, Empty 0; Tup 256419: Vac 0, Keep 0, UnUsed 44822. > Total CPU 0.24s/0.04u sec elapsed 0.30 sec. > NOTICE: Analyzing email_bank_mailing_lists > VACUUM > tradein_clients=# explain SELECT count( email_id ) from email_bank_mailing_lists >where > query_id=499;NOTICE: QUERY PLAN: > > Aggregate (cost=6863.24..6863.24 rows=1 width=4) > -> Seq Scan on email_bank_mailing_lists (cost=0.00..6788.24 rows=30001 width=4) > > EXPLAIN -- Rod Taylor ---------------------------(end of broadcast)--------------------------- TIP 3: if posting/reading through Usenet, please send an appropriate subscribe-nomail command to [EMAIL PROTECTED] so that your message can get through to the mailing list cleanly
