Re: optimizer's cost formulas - Mailing list pgsql-general

From Daniel Bausch
Subject Re: optimizer's cost formulas
Date
Msg-id 51626762.8060700@dvs.tu-darmstadt.de
Whole thread Raw
In response to optimizer's cost formulas  (dafNi <dzaf88@gmail.com>)
Responses Re: optimizer's cost formulas
List pgsql-general
Hi,

AFAIK there is no such thing in the code or documentation.
Nevertheless, for a paper I've published last year, I tried to create
mathematical transcriptions at least for the I/O costs.

http://dl.acm.org/citation.cfm?id=2236584.2236588&coll=DL&dl=ACM&CFID=309621631&CFTOKEN=77712718

Have a look at the appendix.

Regards,
Daniel Bausch

Am 06.04.2013 12:51, schrieb dafNi:
> Hello everybody!
>
> I'm trying to understand the the query planner's cost estimator.
> I was not able to find anywhere the formulas that estimate the cost
> of each operation that the optimizer produces.
>
> I only found this for the sequential scan:
> SEQ SCAN = ( cpu_tuple_cost *  rows ) +  ( number of pages * seq_page_cost )
> where :
> cpu_tuple_cost=0.01
> seq_page_cost=1.0
> and "rows" and "number of pages" are given in the query plan (via
> EXPLAIN ANALYZE)
>
> I am wondering where could I find the rest formulas for the rest operations
> (e.g. HashAggregate, Nested Loop, Hash Join, Index Scan, Sort, etc)
>
> I also looked at costsize.c but could not find a formula like the above
> or maybe I
> couldn't make sense of it
>
>
> Thank you in advance!


--
Daniel Bausch
Wissenschaftlicher Mitarbeiter
Technische Universität Darmstadt
Fachbereich Informatik
Fachgebiet Datenbanken und Verteilte Systeme

Hochschulstraße 10
64289 Darmstadt
Germany

Tel.: +49 6151 16 6706
Fax:  +49 6151 16 6229


pgsql-general by date:

Previous
From: Amit Kapila
Date:
Subject: Re: pg_stat_get_last_vacuum_time(): why non-FULL?
Next
From: Zahid Quadri
Date:
Subject: how to create materialized view in postgresql 8.3