This query runs slowly, and gets slower as the table grows. Can anyone find a way to increase the speed?
It is intended to populate tblUser_Lesson with id's of tblUser and tblLesson while ensuring that the id's don't exist in the table before insert.
I have an index on tblUser.name and tblLesson.name, but it doesn't seem to make a difference.
INSERT INTO tblUser_Lesson (user, lesson)
SELECT userId, lessonId
FROM
(
SELECT tblUser.id userId, tblLesson.id lessonId
FROM tblUser,
tblLesson
WHERE tblUser.name=?
AND tblLesson.name=?
) tmp
WHERE NOT EXISTS (SELECT user
FROM tblUser_Lesson tmp1
WHERE tmp1.user = tmp.userId
AND tmp1.lesson = tmp.tblLesson)
This is an equivalent version of the query that I find easier to read:
INSERT INTO tblUser_Lesson(user, lesson)
SELECT userId, lessonId
FROM tblUser u CROSS JOIN
tblLesson l
WHERE u.name = ? AND l.name = ? AND
NOT EXISTS (SELECT 1
FROM tblUser_Lesson ul
WHERE ul.user = u.userId AND ul.lesson = l.tblLesson
);
My first recommendation is to let the database do the work. Create a unique index on tblUser_Lesson
:
create unique index unq_tblUser_Lesson on tblUser_Lesson(UserId, Lesson);
Then just do the insert as:
INSERT INTO tblUser_Lesson(user, lesson)
SELECT userId, lessonId
FROM tblUser u CROSS JOIN
tblLesson l
WHERE u.name = ? AND l.name = ? ;
Second, I would create indexes for each of the other tables:
create index idx_tbluser_name_id on tblUser(name, id);
create index idx_tblLesson_name_id on tblLesson(name, id);
That speeds this query.
If you (in general) don't want to get an error when there is a duplicate, then you can leave your NOT EXISTS
clause in place. The index on tblUser_Lesson
will still help.