| From: | Kevin Grittner <kgrittn(at)ymail(dot)com> |
|---|---|
| To: | Daniel Begin <jfd553(at)hotmail(dot)com>, 'Bill Moran' <wmoran(at)potentialtech(dot)com> |
| Cc: | "pgsql-general(at)postgresql(dot)org" <pgsql-general(at)postgresql(dot)org>, 'Tomas Vondra' <tomas(dot)vondra(at)2ndquadrant(dot)com>, 'Melvin Davidson' <melvin6925(at)gmail(dot)com> |
| Subject: | Re: Planner cost adjustments |
| Date: | 2015-06-05 13:20:25 |
| Message-ID: | 1084745509.5678555.1433510425269.JavaMail.yahoo@mail.yahoo.com |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-general |
Daniel Begin <jfd553(at)hotmail(dot)com> wrote:
> I can tweak values and restart Postgres without any hardship!
Many of the important performance-related settings (especially cost
factors) can be adjusted with the SET command to affect just the
one connection. This can make experimenting a lot easier.
> About seq_page_cost and random_page_cost, I am about to test
> different lower values as you and Thomas propose.
I have often been unable to get optimal plans without boosting the
value of cpu_tuple_cost (in addition to adjusting the page costs
and setting effective_cache_size). Generally 0.03 is enough,
although I have personally never seen problems with going up to
0.05, and that sometimes fixes a few plans that 0.03 misses.
--
Kevin Grittner
EDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Robert Haas | 2015-06-05 13:39:20 | Re: 9.4.1 -> 9.4.2 problem: could not access status of transaction 1 |
| Previous Message | Thomas Munro | 2015-06-05 10:51:53 | Re: [GENERAL] 9.4.1 -> 9.4.2 problem: could not access status of transaction 1 |