Question

I am doing a query that converts rows to columns similar to this post but have encountered a performance problem. Here is the query:-

SELECT
    Info.Customer,
    Answers.Answer,
    Answers.AnswerDescription,
    Details.Code1,
    Details.Code2,
    Details.Code3
FROM
    Info
    LEFT OUTER JOIN Answers
    ON Info.AnswerID = Answers.AnswerID
    LEFT OUTER JOIN
    (SELECT
      ReferenceNo,
      MAX(CASE DetailsIndicator WHEN 'cde1' THEN DetailsCode ELSE NULL END ) Code1,
      MAX(CASE DetailsIndicator WHEN 'cde2' THEN DetailsCode ELSE NULL END ) Code2,
      MAX(CASE DetailsIndicator WHEN 'cde3' THEN DetailsCode ELSE NULL END ) Code3
    FROM DetailsData
    GROUP BY ReferenceNo) Details
    ON Info.ReferenceNo = Details.ReferenceNo
    WHERE Info.Date BETWEEN x AND y

There are less than 300 rows returned, but the Details table is about 180 thousand rows.
The query takes 45 seconds to run in half the time.
Note the inner query takes 7 seconds to run.

When I type show processlist; into MYSQL it is hanging on “Sending Data”.

Any thoughts as to what the performance problem might be?

Answer 1

First, the inner subselect that queries the “Details” result is querying against ALL entries… is that what you want? I don’t think so. You only appear to want results based on a reference number that qualifies from the outer date check.

So, I would change your inner select to include…

FROM DetailsData 
where DetailsData.ReferenceNo IN 
    ( select distinct Info2.ReferenceNo
        from Info Info2 where Info2.Date between x and y )
GROUP BY ReferenceNo) Details 

This way, you are only getting details associated with reference numbers within the date range in question.