HiveBrain v1.2.0
Get Started
← Back to all entries
patternsqlModerate

Join query taking 11 mins to run on 300,000 rows table

Submitted by: @import:stackexchange-dba··
0
Viewed 0 times
300rows000queryjoinminstableruntaking

Problem

Below query is taking more than 11 minutes to execute.

SELECT `c`.*, 
       `e`.`name`                                                 AS `employee_name`, 
       `e`.`emp_no`, 
       `d`.`code`                                                 AS `department_code`, 
       IF(ew.code IS NOT NULL, ew.code, egw.code)                 AS shift_code, 
       IF(ew.code IS NOT NULL, ew.time_in_from, egw.time_in_from) AS time_in_from, 
       IF(ew.code IS NOT NULL, ew.time_out_to, egw.time_out_to)   AS time_out_to, 
       IF(ew.code IS NOT NULL, ew.next_day, egw.next_day)         AS next_day 
FROM   `tms_emp_badge_card` AS `c` 
       LEFT JOIN `tms_door_record_raw` AS `dr` 
              ON `c`.`card_no` = `dr`.`card_no` 
       LEFT JOIN `tms_employee` AS `e` 
              ON `c`.`emp_no` = `e`.`emp_no` 
       LEFT JOIN `tms_emp_group` AS `g` 
              ON `e`.`group_id` = `g`.`id` 
       LEFT JOIN `tms_emp_department` AS `d` 
              ON `e`.`department_id` = `d`.`id` 
       LEFT JOIN `tms_emp_workschedule` AS `ewt` 
              ON `ewt`.`workschedule_date` = "2016-11-01" 
                 AND ( ewt.reference_no = c.emp_no 
                       AND ewt.reference_type = "emp_no" ) 
       LEFT JOIN `tms_workschedule` AS `ew` 
              ON `ewt`.`workschedule_id` = `ew`.`id` 
       LEFT JOIN `tms_emp_workschedule` AS `egwt` 
              ON `egwt`.`workschedule_date` = "2016-11-01" 
                 AND ( egwt.reference_no = g.code 
                       AND egwt.reference_type = "group_code" ) 
       LEFT JOIN `tms_workschedule` AS `egw` 
              ON `egwt`.`workschedule_id` = `egw`.`id` 
WHERE  `dr`.`record_time` BETWEEN '2016-11-01' AND '2016-11-02' 
GROUP  BY `c`.`card_no` 
ORDER  BY c.emp_no


Below is the explain query

```
+----+-------------+-------+------------+--------+-------------------------------------------------------------------------+-----------------------------------------+---------+--------------------------+-

Solution

The following should help with the execution time:

  • remove the ORDER BY if it isn't strictly necessary



  • replace the join of dr table with WHERE EXISTS (SELECT 1 FROM tms_door_record_raw As dr WHERE c.card_no = dr.card_no AND dr.record_time BETWEEN '2016-11-01' AND '2016-11-02')



  • GROUP BY, may now not be needed



  • widen the index on tms_door_record_raw to include both card_no and record_time



Test this and see if progress is being made. Further steps may be necessary but hopefully this is in the right direction.

Context

StackExchange Database Administrators Q#156853, answer score: 11

Revisions (0)

No revisions yet.