Re: Analyze tool?
От | Thom Brown |
---|---|
Тема | Re: Analyze tool? |
Дата | |
Msg-id | AANLkTi=HRnmxT_pH1a87rq6+Or-juPSFzVWZz0pOnZ_N@mail.gmail.com обсуждение исходный текст |
Ответ на | Re: Analyze tool? (Bjørn T Johansen <btj@havleik.no>) |
Список | pgsql-general |
2010/10/1 Bjørn T Johansen <btj@havleik.no>: > On Fri, 1 Oct 2010 11:30:59 +0100 > Thom Brown <thom@linux.com> wrote: > >> 2010/10/1 Bjørn T Johansen <btj@havleik.no>: >> > We are using both DB2 and PostgreSQL at work and DB2 has a nice tool, i5 Navigator, where one can enable logging ofSQL statements and then it >> > will recommed indexes that should/could be created to increase speed... >> > Does there exist a similar tool for PostgreSQL? >> >> You can set log_min_duration_statement to log statements which take >> over a certain amount of time, and then use pgFouine to read the log >> files and identify the most frequently run queries, and the longest >> queries. > > Sounds like a something that should be tried... > >> >> You can also use the auto_explain contrib module >> (http://www.postgresql.org/docs/9.0/static/auto-explain.html) to log >> the plans of queries which take too long. However, I don't think >> pgFouine can use those outputs.. at least not yet. > > Ok, plan B... > >> >> But to find out what indexes you'll need, getting used to reading >> query plans will help as it will show you more than just where >> sequentials scans are taking place. It will also show you what the >> planner believes a query will cost compared to how much it actually >> costs, which can provide insight into tables which require vacuuming, >> indexes which might need clustering, or table stats which require >> modifying to match you data. > > Yes, but it would be nice to be pointed in the right direction first, but it seems like log_min_duration_statement canbe used for that... > And also running explain involves manually work, would have been nice with some automatic procedure.... I agree. I believe MSSQL has a similar tool too, so surprised a 3rd party hasn't already come up with one. -- Thom Brown Twitter: @darkixion IRC (freenode): dark_ixion Registered Linux user: #516935
В списке pgsql-general по дате отправления: