Re: What does it mean? Plan stats and double rainbows.
| От | David G. Johnston |
|---|---|
| Тема | Re: What does it mean? Plan stats and double rainbows. |
| Дата | |
| Msg-id | CAKFQuwYzwDByK7g6BoWvR6DeAqW5wuGff5QCrSAEkB8g8KT9ww@mail.gmail.com обсуждение исходный текст |
| Ответ на | What does it mean? Plan stats and double rainbows. (Michael Moore <michaeljmoore@gmail.com>) |
| Ответы |
Re: What does it mean? Plan stats and double rainbows.
|
| Список | pgsql-sql |
I'm having a difficult time finding documentation on EXPLAIN PLAN stats. For example, in
' -> Nested Loop Left Join (cost=0.43..1415.06 rows=2 width=1377) (actual time=0.093..0.093 rows=0 loops=1)'
what does 0.43..1415.06 mean? Is that a range? If so, it seems rather pointless, like saying "somewhere between 0 and infinity".
Thomas' link should cover this but it isn't giving you a probabilistic range , its giving the time to first record and time to fetch all records. For stuff like semi-joins you don't care about the total number of records found only that you can quickly find one record. A limited requirement but since plan output is somewhat generic in nature it always gives both numbers.
Also, is there a way to tell the query planner to limit the search for the best plan on a per statement basis. I know that this exists as a config parameter but I think that applies to the entire database. I have a query that takes 9 times more time to plan than it does to execute.
All parameters (in this context) are session-local in use; even if the default value is set at the scope of the entire server. You can make them transaction-local by using "SET LOCAL" instead of a plain "SET" when changing them within the session.
David J.
В списке pgsql-sql по дате отправления: