mysql derived2、Using temporary 和 Using filesort 优化记录

-- 业务每周查询一份报表，每次执行sql磁盘直接由49%涨到%93，执行了几次发现是sql用到了磁盘临时表，对sql进行了改写
-- 原查询sql：

SELECTCODE,NAME,num,GROUP_CONCAT(product_id) AS product_ids,GROUP_CONCAT(product_name) AS product_names
FROM(SELECTCOUNT(DISTINCT (psn.Psnid)) AS num,cus.CODE,cus.`name`,od.product_id,pr.NAME AS product_nameFROMmid_psn_copy AS psnJOIN customer AS cus ON cus.CODE = psn.PkCorpJOIN product_order_rel AS od ON od.customer_code = cus.CODEJOIN product AS pr ON pr.id = od.product_idWHEREpsn.`Status` = 1 AND od.`status` = 1GROUP BY od.customer_code , od.product_id) AS tt
GROUP BY CODE;-- 69 rows in set, 1 warning (34.54 sec)

-- 原执行计划

+----+-------------+------------+--------+---------------------------------------------------------------------------+------------------------------+---------+------------------------+--------+----------------------------------------------+
| id | select_type | table      | type   | possible_keys                                                             | key                          | key_len | ref                    | rows   | Extra                                        |
+----+-------------+------------+--------+---------------------------------------------------------------------------+------------------------------+---------+------------------------+--------+----------------------------------------------+
|  1 | PRIMARY     | <derived2> | ALL    | NULL                                                                      | NULL                         | NULL    | NULL                   | 325296 | Using filesort                               |
|  2 | DERIVED     | cus        | index  | idx_custom_code,idx_code_name                                             | idx_code_name                | 274     | NULL                   |    251 | Using index; Using temporary; Using filesort |
|  2 | DERIVED     | psn        | ref    | idx_psn_pkcorp,idx_psn_Status,idx_pkcrop_status                           | idx_pkcrop_status            | 157     | gmcloud.cus.code,const |     36 | Using index condition                        |
|  2 | DERIVED     | od         | ref    | idx_customer_code_pdid_status,idx_product_id,idx_customer_code_product_id | idx_customer_code_product_id | 123     | gmcloud.cus.code       |     36 | Using where; Using index                     |
|  2 | DERIVED     | pr         | eq_ref | PRIMARY                                                                   | PRIMARY                      | 8       | gmcloud.od.product_id  |      1 | NULL                                         |
+----+-------------+------------+--------+---------------------------------------------------------------------------+------------------------------+---------+------------------------+--------+----------------------------------------------+

-- 第一次优化，去掉外层嵌套，把外层条件融合到当前sql
-- 加联合索引
-- alter table customer add key idx_code_name (`code`,`name`)
-- alter table product_order_rel add key idx_customer_code_pdid_status (`customer_code`,`product_id`,`status`)

SELECTCOUNT(DISTINCT (psn.Psnid)) AS num,cus.CODE,cus.`name`,GROUP_CONCAT(od.product_id) AS product_ids,GROUP_CONCAT(pr.NAME) AS product_namesFROMmid_psn_copy AS psnJOIN customer AS cus ON cus.CODE = psn.PkCorpJOIN product_order_rel AS od ON od.customer_code = cus.CODEJOIN product AS pr ON pr.id = od.product_idWHEREpsn.`Status` = 1 AND od.`status` = 1GROUP BY cus.CODE, od.customer_code , od.product_id;

-- 查看执行计划，虽然没有了DERIVED、derived2，但是看到还有Using temporary; Using filesort存在

+----+-------------+-------+--------+-------------------------------+-------------------------------+---------+-----------------------+------+----------------------------------------------+
| id | select_type | table | type   | possible_keys                 | key                           | key_len | ref                   | rows | Extra                                        |
+----+-------------+-------+--------+-------------------------------+-------------------------------+---------+-----------------------+------+----------------------------------------------+
|  1 | SIMPLE      | cus   | index  | idx_custom_code,idx_code_name | idx_code_name                 | 274     | NULL                  |  251 | Using index; Using temporary; Using filesort |
|  1 | SIMPLE      | psn   | ref    | idx_psn_pkcorp,idx_psn_Status | idx_psn_pkcorp                | 152     | gmcloud.cus.code      |   69 | Using index condition; Using where           |
|  1 | SIMPLE      | od    | ref    | idx_customer_code_pdid_status | idx_customer_code_pdid_status | 123     | gmcloud.cus.code      |   36 | Using where; Using index                     |
|  1 | SIMPLE      | pr    | eq_ref | PRIMARY                       | PRIMARY                       | 8       | gmcloud.od.product_id |    1 | NULL                                         |
+----+-------------+-------+--------+-------------------------------+-------------------------------+---------+-----------------------+------+----------------------------------------------+

-- 第二次优化，把group by 后边的od.customer_code , od.product_id去掉，在前边group_concat加上distinct
-- 在where条件和group by条件都加了联合索引
-- alter table customer add key idx_code_name (`code`,`name`)
-- alter table mid_psn_copy add key idx_pkcrop_status` (`PkCorp`,`Status`)
-- alter table product_order_rel add key idx_customer_code_product_id (`customer_code`,`product_id`,`status`)
-- 优化完的sql：

SELECTCOUNT(DISTINCT (psn.Psnid)) AS num,cus.CODE,cus.`name`,GROUP_CONCAT(distinct od.product_id) AS product_ids,GROUP_CONCAT(distinct pr.NAME) AS product_namesFROMcustomer AS cusJOIN product_order_rel AS od ON od.customer_code = cus.CODEJOIN mid_psn_copy AS psn ON  psn.PkCorp = cus.CODEJOIN product AS pr ON pr.id = od.product_idWHEREpsn.`Status` = 1 AND od.`status` = 1GROUP BY cus.CODE;

-- 再次查看执行计划

+----+-------------+-------+--------+-------------------------------------------------+------------------------------+---------+------------------------+------+--------------------------+
| id | select_type | table | type   | possible_keys                                   | key                          | key_len | ref                    | rows | Extra                    |
+----+-------------+-------+--------+-------------------------------------------------+------------------------------+---------+------------------------+------+--------------------------+
|  1 | SIMPLE      | cus   | index  | idx_code_name,idx_code                          | idx_code_name                | 274     | NULL                   |  251 | Using index              |
|  1 | SIMPLE      | psn   | ref    | idx_psn_pkcorp,idx_psn_Status,idx_pkcrop_status | idx_pkcrop_status            | 157     | gmcloud.cus.code,const |   36 | Using index condition    |
|  1 | SIMPLE      | od    | ref    | idx_product_id,idx_customer_code_product_id     | idx_customer_code_product_id | 123     | gmcloud.cus.code       |   36 | Using where; Using index |
|  1 | SIMPLE      | pr    | eq_ref | PRIMARY                                         | PRIMARY                      | 8       | gmcloud.od.product_id  |    1 | NULL                     |
+----+-------------+-------+--------+-------------------------------------------------+------------------------------+---------+------------------------+------+--------------------------+

-- 调整下表头顺序

SELECTcus.CODE,cus.`name`,COUNT(DISTINCT (psn.Psnid)) AS num,GROUP_CONCAT(DISTINCT od.product_id) AS product_ids,GROUP_CONCAT(DISTINCT pr.NAME) AS product_names
FROMcustomer AS cusJOINproduct_order_rel AS od ON od.customer_code = cus.CODEJOINmid_psn_copy AS psn ON psn.PkCorp = cus.CODEJOINproduct AS pr ON pr.id = od.product_id
WHEREpsn.`Status` = 1 AND od.`status` = 1
GROUP BY cus.CODE;
-- 69 rows in set (6.95 sec)

-- 执行sql后，查看临时表状态没有增加了
mysql> show status like '%tmp%';

+-------------------------+-------+
| Variable_name           | Value |
+-------------------------+-------+
| Created_tmp_disk_tables | 0     |
| Created_tmp_files       | 619   |
| Created_tmp_tables      | 3     |
+-------------------------+-------+

优化待续。。。

mysql derived2、Using temporary 和 Using filesort 优化记录相关推荐

mysql优化-oder by产生的Using temporary与Using filesort问题解决
下面的mysql代码因为order by的原因,出现了Using temporary与Using filesort,优化前代码如下 SELECT DISTINCT( tm.task_no ) task ...
mysql using temporary_执行计划中Using filesort,Using temporary相关语句的优化解决_MySQL
bitsCN.com 昨天听开发人员提到,相关的彩票网页当中一个页面刷新的很慢,特别是在提取数据的时候, 今天早上一到,便去找开发人员要去相关的也没进行浏览,窥探哪些数据出现了问题,开发人员使用PH ...
MySQL 的索引是什么？怎么优化？
转载自 MySQL 的索引是什么?怎么优化? 摘要: 索引对大数据的查询速度的提升是非常大的,Explain可以帮你分析SQL语句是否用到相关索引. 索引类似大学图书馆建书目索引,可以提高数据检索的 ...
我的MYSQL学习心得（十六）优化
原文:我的MYSQL学习心得(十六) 优化我的MYSQL学习心得(十六) 优化我的MYSQL学习心得(一) 简单语法我的MYSQL学习心得(二) 数据类型宽度我的MYSQL学习心得(三) 查看 ...
mybatis与mysql调优_MySQL + mybatis的SQL优化方案
sql优化方案: 1.添加索引,在条件参数,关联参数上建立参数, 2.字段优化,需要什么字段查什么字段 3.模糊查询尽量使用: select * from tableName a where a.na ...
用MySql的查询分析语法explain来优化查询和索引
http://hi.baidu.com/wtnzone/item/beb83840a4971af4dd0f6c77 数据库最常见的操作就是查询了,我们经常要用"SELECT"语法对 ...
MySQL 的索引、执行计划、优化器算法
SQL处理流程 INDEX 索引索引介绍索引:是排序的快速查找的特殊数据结构,定义作为查找条件的字段上,又称为键key,索引通过存储引擎实现: 索引相当于一本书的目录,可以优化查询. 优点: 索引 ...
MySQL的批量更新和批量新增优化
MySQL的批量更新和批量插入优化如果需要批量插入和批量更新操作就需要进行sql 的优化,否则近30万条数据的插入或更新就会耗费几分钟甚至更多的时间, 此文仅批量插入和批量更新的几种优化. 批量插入 ...
Mysql中慢SQL的分析与优化
为何对慢SQL进行治理从数据库角度看:每个SQL执行都需要消耗一定I/O资源,SQL执行的快慢,决定资源被占用时间的长短.假设总资源是100,有一条慢SQL占用了30的资源共计1分钟.那么在这1分钟 ...

mysql derived2、Using temporary 和 Using filesort 优化记录

mysql derived2、Using temporary 和 Using filesort 优化记录相关推荐

最新文章

热门文章