Horton

Re: BRIN cost estimate

Emre Hasegeli <emre@hasegeli.com>

View thread View on postgr.es
From: Emre Hasegeli <emre@hasegeli.com>
To: David Rowley <david.rowley@2ndquadrant.com>
Cc: Alvaro Herrera <alvherre@2ndquadrant.com>, PostgreSQL Hackers <pgsql-hackers@postgresql.org>
Date: 2017-04-05T05:34:29Z
Lists: pgsql-hackers 
> Interested to hear comments on this.


I don't have chance to test it right now, but I am sure it would be an
improvement over what we have right now.  There is no single correct
equation with so many unknowns we have.

>   *indexTotalCost += (numTuples * *indexSelectivity) *
(cpu_index_tuple_cost + qual_op_cost);

Have you preserved this line intentionally?  This would make BRIN index
scan cost even higher, but the primary property of BRIN is to be cheap to
scan. Shouldn't bitmap heap scan node take into account of checking the
tuples in its cost?

Commits

  1. Fix BRIN cost estimation

  2. Make use of statistics on index expressions. There are still some