问题
I have a preparedstatement from a framework as followed:
SELECT OH.ORDER_ID, MAX(OS.STATUS_DATETIME) FROM public.ORDER_HEADER OH, public.ORDER_STATUS OS WHERE ((OH.ORDER_ID = OS.ORDER_ID AND OH.STATUS_ID = ? AND OS.STATUS_ID = ?)) GROUP BY OH.ORDER_
ID HAVING (MAX(OS.STATUS_DATETIME) <= ?) ORDER BY OH.ORDER_ID ASC
OrderHeader has 300 000 rows, OrderStatus has roughly 5-6 million. All queried fields have an index on them. The database is Postgres 9.1
SELECT OH.ORDER_ID, MAX(OS.STATUS_DATETIME) FROM ORDER_HEADER OH INNER JOIN ORDER_STATUS OS
ON ((OH.ORDER_ID = OS.ORDER_ID AND OH.STATUS_ID = 'ORDER_PARTIALLY_RECEIVED' AND OS.STATUS_ID = 'ORDER_PARTIALLY_RECEIVED'))
GROUP BY OH.ORDER_ID HAVING (MAX(OS.STATUS_DATETIME) <= '2015-01-27 00:00:00') ORDER BY OH.ORDER_ID ASC
This is the query string when the parameters are substitued in.
Running the query from Java, as a PreparedStatement and parameters set in there is making the query to run for minutes (5-6 minutes). When i run the SQL manually, it takes 10 seconds. How is this possible, how can i explain it?
回答1:
Prepared statements are optimized without knowledge of actual parameters. In PostgreSQL 9.1 a prepared statements are optimized only for most common values. When prepared statements is slow, then you have to use a dynamic SQL. PostgreSQL has nice API - parametric queries - it is some between prepared statements and usual queries - It is safe against SQL injection, and it is immune against a problems with blind optimization.
回答2:
If the result set is large - PreparedStatement could work slower, because default "fetchSize" is only 10.
Try to set bigger fetchSize for this query.
More info here: http://docs.oracle.com/cd/A97335_02/apps.102/a83724/resltse5.htm
And specialy for Postgres here: https://jdbc.postgresql.org/documentation/head/query.html
回答3:
If you use preparedStatements,you must identify paramter type . you can explain the sql.
e.g
uid type is integer
explain analyse select * from users where uid in (200,500,800);
PREPARE fooplan (int, int, int) AS
select *
from users
where uid in ($1, $2, $3);
explain analyse EXECUTE fooplan(200, 500, 800);
PREPARE fooplan1 (numeric, numeric, numeric) AS
select *
from users
where uid in ($1, $2, $3);
explain analyse EXECUTE fooplan1(200, 500, 800);
来源:https://stackoverflow.com/questions/28236827/preparedstatement-very-slow-but-manual-query-quick