Re: NL Join vs Merge Join - 5 hours vs 2 seconds

2018-12-17 Thread David Rowley
On Tue, 18 Dec 2018 at 09:21, Kumar, Virendra wrote: > I tried query with function dependency statistics but it did not help. I > don't see we can create such statistics on table join. There's mention of why this is the case in the docs in [1]: "14.2.2.1.1. Limitations of Functional Dependencie

RE: NL Join vs Merge Join - 5 hours vs 2 seconds

2018-12-17 Thread Kumar, Virendra
02 PM To: Kumar, Virendra Cc: pgsql-general@lists.postgresql.org Subject: Re: NL Join vs Merge Join - 5 hours vs 2 seconds "Kumar, Virendra" writes: > We are using Postgers-XL based on PostgreSQL 10.5 on RHEL7. We have an > extreme case of one query which running in 4 hour 45 min

Re: NL Join vs Merge Join - 5 hours vs 2 seconds

2018-12-17 Thread Tom Lane
"Kumar, Virendra" writes: > We are using Postgers-XL based on PostgreSQL 10.5 on RHEL7. We have an > extreme case of one query which running in 4 hour 45 mins and 33 seconds with > Nested Loops Join vs 2.5 seconds when we disable enable_nestloop. The query > is given below. I would be happy to