mysql查询执行耗时

Question

I have seen several question in SO and based in that I improved my sql query also. but it sometime take 12 second or it sometime takes 3 seconds to execute. so minimum time we can its 3 seconds. query is like this way

SELECT ANALYSIS.DEPARTMENT_ID
    ,SCORE.ID
    ,SCORE.KPI_ SCORE.R_SCORE
    ,SCORE.FACTOR_SCORE
    ,SCORE.FACTOR_SCORE
    ,SCORE.FACTOR_SCORE
    ,SCORE.CREATED_DATE
    ,SCORE.UPDATED_DATE
FROM SCORE_INDICATOR SCORE
    ,AG_SENTIMENT ANALYSIS
WHERE SCORE.TAG_ID = ANALYSIS.ID
    AND ANALYSIS.ORGANIZATION_ID = 1
    AND ANALYSIS.DEPARTMENT_ID IN (1,2,3,4,5)
    AND DATE (ANALYSIS.REVIEW_DATE) BETWEEN DATE ('2016-05-02') AND DATE ('2017-05-02')
ORDER BY ANALYSIS.DEPARTMENT_ID

now one table SCORE_INDIACATOR has 19345116 and later has 19057025 rows total. and I added index on ORGANIZATION_ID and department_id and another as combination of ORGANIZATION_ID and department_id . is there any other way to improve it or is it maximum I can achieve with this amount of data?

Answer 1

Here is checklist:

1) Make sure logs table ( ANALYSIS ) uses MyISAM engine (it's fast for OLAP queries).

2) Make sure that You've indexed ANALYSIS.REVIEW_DATE field.

3) Make sure that ANALYSIS.REVIEW_DATE is type of DATE (not CHAR , VARCHAR )

4) Change query (rearrange query plan):

SELECT 
    ANALYSIS.DEPARTMENT_ID
    ,SCORE.ID
    ,SCORE.KPI_ SCORE.R_SCORE
    ,SCORE.FACTOR_SCORE
    ,SCORE.FACTOR_SCORE
    ,SCORE.FACTOR_SCORE
    ,SCORE.CREATED_DATE
    ,SCORE.UPDATED_DATE
FROM SCORE_INDICATOR SCORE
    ,AG_SENTIMENT ANALYSIS
WHERE 
    SCORE.TAG_ID = ANALYSIS.ID 
  AND
    ANALYSIS.REVIEW_DATE >= '2016-05-02' AND ANALYSIS.REVIEW_DATE < '2016-05-03'
  AND
    ANALYSIS.ORGANIZATION_ID = 1
  AND 
    ANALYSIS.DEPARTMENT_ID IN (1,2,3,4,5)
ORDER BY ANALYSIS.DEPARTMENT_ID;

Answer 2

I have changed the order and style to JOIN syntax. The Score table seems to be the child to the primary criteria of the Analysis table. All your criteria is based on qualifying Analysis records. Now, the indexing. By doing a DATE() function call on a column does not help the optimizer. So, to get all possible date/time components, I have changed from between to >= the first date and LESS THAN one day beyond the end. In your example DATE( '2017-05-02' ) is the same as LESS than '2017-05-03' which will include 2017-05-02 up to 23:59:59 and the date can be applied better.

Now for the index. DO a compound index based on fields for join and order by might help

AG_Segment table... index ON(Organization_ID, Department_ID, Review_Date, ID)

SELECT 
        ANALYSIS.DEPARTMENT_ID,
        SCORE.ID,
        SCORE.KPI_ SCORE.R_SCORE,
        SCORE.FACTOR_SCORE,
        SCORE.FACTOR_SCORE,
        SCORE.FACTOR_SCORE,
        SCORE.CREATED_DATE,
        SCORE.UPDATED_DATE
    FROM 
        AG_SENTIMENT ANALYSIS
            JOIN SCORE_INDICATOR SCORE
                ON ANALYSIS.ID = SCORE.TAG_ID
    where 
            ANALYSIS.ORGANIZATION_ID = 1
        AND ANALYSIS.DEPARTMENT_ID IN (1,2,3,4,5)
        AND ANALYSIS.REVIEW_DATE >= '2016-05-02'
        AND ANALYSIS.REVIEW_DATE < '2017-05-03'
    ORDER BY 
        ANALYSIS.DEPARTMENT_ID