"Stephen Denne" <Stephen.Denne@xxxxxxxxxxxxxx> writes: > Pallav Kalva asked > ... >> and listing0_.fklistingsourceid=5525 > ... >> order by listing0_.entrydate desc limit 10; > >> -> Index Scan Backward using idx_listing_entrydate on >> listing listing0_ (cost=0.00..781557.28 rows=5118 width=107) (actual >> time=2113544.412..2113544.412 rows=0 loops=1) >> Filter: (fklistingsourceid = 5525) > > Would it help to have a combined index on fklistingsourceid, entrydate? I think that would help. You already have a ton of indexes, you might consider whether all your queries start with a listingsourceid and whether you can have that as a prefix on the existing index. Another thing to try is raising the stats target on fklistingsourceid and/or entrydate. The estimate seems pretty poor. It could just be that the distribution is highly skewed which is a hard case to estimate correctly. -- Gregory Stark EnterpriseDB http://www.enterprisedb.com Ask me about EnterpriseDB's RemoteDBA services! ---------------------------(end of broadcast)--------------------------- TIP 7: You can help support the PostgreSQL project by donating at http://www.postgresql.org/about/donate
