Thanks Tom, I think you're right. I just did an analyze on table test1 and the execution plan now generated is more stable and predictable. Thanks, Suya -----Original Message----- From: Tom Lane [mailto:tgl@xxxxxxxxxxxxx] Sent: Tuesday, May 20, 2014 12:22 AM To: Huang, Suya Cc: pgsql-performance@xxxxxxxxxxxxxx Subject: Re: [PERFORM] same query different execution plan (hash join vs. semi-hash join) "Huang, Suya" <Suya.Huang@xxxxxxxxxxxxxxx> writes: > Thank you Tom. But the time spent on scanning table test1 is less than 1 second (91.738 compares to 87.869), so I guess this shouldn't be the issue? No, the point is that the bad rowcount estimate (and, possibly, lack of stats about join column contents) causes the planner to pick a join method that's not ideal for this query. regards, tom lane
