Re: Request improve pg_stat_statements module
Robert Haas <robertmhaas@gmail.com>
From: Robert Haas <robertmhaas@gmail.com>
To: Michael Paquier <michael.paquier@gmail.com>
Cc: pgsql-kr@postgresql.kr,
PostgreSQL mailing lists <pgsql-hackers@postgresql.org>
Date: 2014-03-03T15:36:35Z
Lists: pgsql-hackers
On Fri, Feb 28, 2014 at 8:01 AM, Michael Paquier <michael.paquier@gmail.com> wrote: > Thanks for your patch! > > On Fri, Feb 28, 2014 at 4:18 PM, <pgsql-kr@postgresql.kr> wrote: >> I patched to add one column in pg_stat_statements module. >> and sent to author but >> I need a last time of query, because I want to analyse order by recent time. > Hm... I am not sure that this brings much to pg_stat_statements, which > is interested to gather normalized information about the queries run > on the server. For example, things like calculating the average time > of a query by using total_time/calls or even the total time to guess > where is an application bottleneck is interesting on a busy system, > while the latest timestamp is not really an information that can be > used for query tuning. Getting the latest timestamp when a query has > run is particularly not interesting on OLTP-like applications where > many short transactions are running. It is more interesting to > classify query results by either calls, total_time or the combination > of both IMO. FWIW, I think it's a neat idea, but I think this proposal is probably too late for 9.4. -- Robert Haas EnterpriseDB: http://www.enterprisedb.com The Enterprise PostgreSQL Company