Link to home
Start Free TrialLog in
Avatar of hnmcc
hnmcc

asked on

Very slow MySQL query

I am writing a fairly simple web application in PHP which sits on a pre-existing MySQL database which belongs to my client.  It is to replace an existing application which will soon shut down.

One of the queries runs very slowly indeed - I haven't benchmarked it, but we're talking seconds, a really noticeable gap.  But the existing application, which works in a very similar way, does not have this delay.

I know nothing about optimising SQL queries (in MySQL or anywhere else).  Can anyone help me?

The query is:

SELECT `fid`, `enid`, `label`, `activityname`, `date`, `aid_FK`, `extension`
FROM first INNER JOIN second INNER JOIN third
ON first.fid = second.fid_FK AND second.aid_FK = third.aid
WHERE `mem_no_FK` = $var
ORDER BY `date`

Details of the tables are attached as a file.

It may be relevant that table 'first' holds user files - WP documents, spreadsheets, etc. - in its `data` field.

question-1.txt
Avatar of Guy Hengel [angelIII / a3]
Guy Hengel [angelIII / a3]
Flag of Luxembourg image

can you post the EXPLAIN plan for the query, please?
Avatar of hnmcc
hnmcc

ASKER

EXPLAIN plan attached (rows don't line up here). question-2.txt
>WHERE `mem_no_FK` = 2353
do you have an index on that field (presuming it's from "first" table)?
if not: create one, and repost the explain plan then (if you are not happy with the new execution time)

also, I suggest you change the SQL a bit to increase readablity (and eventually query execution):
* add table aliases to your tables, and use them on ALL the columns (see the code below, I had to guess, though)
* put the JOIN conditions just below/beside the actual join
SELECT f.`fid`, `f.enid`, s.`label`, s.`activityname`, f.`date`, s.`aid_FK`, t.`extension`
FROM first f
INNER JOIN second s
  ON f.fid = s.fid_FK 
INNER JOIN third t
  ON s.aid_FK = t.aid
WHERE f.`mem_no_FK` = $var
ORDER BY f.`date`

Open in new window

Avatar of hnmcc

ASKER

OK (sorry for delay in responding - I've been away).  I've improved the naming for clarity's sake, and stripped away all the fake names (too boring...).  This is the query:

SELECT f.`fid`, f.`label`, f.`extension`, u.`enid`, a.`activityname`, a.`date` FROM files f JOIN user_entries u ON f.`fid` = u.`fid_FK` JOIN activities a ON u.`aid_FK` = a.`aid` WHERE f.`mem_no_FK` = $var ORDER BY a.`date`

Open in new window


There is a key in files on `mem_no_FK` (B-tree, not unique).

As the rows in files are each in a 1-to-1 relationship with a row in user_entries, would it be better to insert the contents of files into user_entries?  All rows in files have a corresponding row in user_entries, but most rows in user_entries have no corresponding row in files.

Unfortunately, after adding the necessary columns to a copy of user_entries, my attempt to copy the data across fails:
#1054 - Unknown column 'user_entries_x.fid_FK' in 'where clause'
INSERT INTO user_entries_x (`filename`, `extension`, `data`, `size`, `uploaded`, `edited`, `label`) SELECT `filename`, `extension`, `data`, `size`, `created`, `modified`, `label` FROM files WHERE files.`fid` = user_entries_x.`fid_FK`

Open in new window

Avatar of hnmcc

ASKER

EXPLAIN plan attached. SQL-result---phpMyAdmin-3-3-2deb.pdf
ASKER CERTIFIED SOLUTION
Avatar of Guy Hengel [angelIII / a3]
Guy Hengel [angelIII / a3]
Flag of Luxembourg image

Link to home
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Start Free Trial
Avatar of hnmcc

ASKER

There is (and has been since I inherited the database) an index on files.mem_no_FK.  I've now added one to user_entries.fid_fk as you suggest, and checked the join fields for compatibility.  They were all the same data types, but in each case the original was unsigned and the matching FK wasn't.  Does that matter? I changed them anyway.

Altogether this improves the query's performance considerably: c.15s to 0.2s - which is good enough for the requirement.

Thanks for your help.