ITPub博客

首页 > 数据库 > MySQL > MySQL的SQL语句优化一例

MySQL的SQL语句优化一例

原创 MySQL 作者:甲骨文技术支持 时间:2017-02-14 13:54:59 0 删除 编辑

今天在系统上看到一条sql运行时间达到9秒,不符合规范要求,优化之~~

sql如下:sql强制用了into_time索引


  1. # Time: 2017-02-14T11:35:01.594499+08:00
  2. # User@Host: oms_readonly[oms_readonly] @ [10.44.xxx.xxx] Id: 41636892
  3. # Query_time: 9.299612 Lock_time: 0.000124 Rows_sent: 20 Rows_examined: 2439330
  4. SET timestamp=1487043301;
  5. select * from customers force index(`into_time`) where `type` = 1 AND `status` < 7 AND `isarea` = 8 AND `into_time`<='2017-01-31 23:59:59' order by score desc limit 0,20;
查看表的相关状态:


  1. mysql> show table status like 'customers' \G;
  2. *************************** 1. row ***************************
  3.            Name: customers
  4.          Engine: InnoDB
  5.         Version: 10
  6.      Row_format: Dynamic
  7.            Rows: 2504609
  8.  Avg_row_length: 710
  9.     Data_length: 1780383744
  10. Max_data_length: 0
  11.    Index_length: 1253048320
  12.       Data_free: 6291456
  13.  Auto_increment: 2546101
  14.     Create_time: 2017-01-07 01:59:34
  15.     Update_time: 2017-02-14 13:58:17
  16.      Check_time: NULL
  17.       Collation: utf8_general_ci
  18.        Checksum: NULL
  19.  Create_options:
  20.         Comment:
  21. 1 row in set (0.00 sec)
表一共大约有250万行记录,查看下满足into_time<='2017-01-31 23:59:59'这个条件的有多少行


  1. mysql> select count(*) from customers where `into_time`<='2017-01-31 23:59:59';
  2. +----------+
  3. | count(*) |
  4. +----------+
  5. | 2439147 |
  6. +----------+
  7. 1 row in set (0.95 sec)

显然into_time这个列的索引已经不合适了,查看下表上都有那些索引


  1. Create Table: CREATE TABLE `customers` (
  2.   `id` int(11) NOT NULL AUTO_INCREMENT,
  3.   PRIMARY KEY (`id`),
  4.   KEY `newdata` (`newdata`),
  5.   KEY `cusname` (`cusname`),
  6.   KEY `type` (`type`,`ownerid`),
  7.   KEY `operator` (`operator`),
  8.   KEY `into_time` (`into_time`),
  9.   KEY `isarea` (`isarea`),
  10.   KEY `linkcase` (`linkcase`),
  11.   KEY `score` (`score`),
  12.   FULLTEXT KEY `fdx_cusname` (`cusname_idx`)
  13. ) ENGINE=InnoDB AUTO_INCREMENT=2546101 DEFAULT CHARSET=utf8
  14. 1 row in set (0.00 sec)
可以看到score列有索引,如果能采用这个列的索引是个比较好的选择,去掉强制索引看下执行计划



    1. mysql> explain select * from customers where `type` = 1 AND `status` < 7 AND `isarea` = 8 AND `into_time`<='2017-01-31 23:59:59' order by score desc limit 0,20;
    2. +----+-------------+-----------+------------+-------+-----------------------------------------------------------------+-------+---------+------+------+----------+-------------+
    3. | id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
    4. +----+-------------+-----------+------------+-------+-----------------------------------------------------------------+-------+---------+------+------+----------+-------------+
    5. | 1 | SIMPLE | customers | NULL | index | type,into_time,isarea,status,idx_isarea_renew_owner,type_status | score | 2 | NULL | 270 | 0.92 | Using where |
    6. +----+-------------+-----------+------------+-------+-----------------------------------------------------------------+-------+---------+------+------+----------+-------------+
    7. 1 row in set, 1 warning (0.00 sec)


可以看到用了score索引,执行时间从最9秒多,到优化后的0.0几秒。


来自 “ ITPUB博客 ” ,链接:https://blog.itpub.net/20893244/viewspace-2133512/ ,如需转载,请注明出处,否则将追究法律责任。

请登录后发表评论 登录
全部评论
2009年获得Oracle OCM认证,ITPUB社区OCM联盟成员,熟悉Oracle,mysql,Python,Zabbix等。

注册时间:2016-12-31

  • 博文量
    143
  • 访问量
    570503