为某XXX手机公司删除重复数据优化一例

这是他们一个开发写的SQL，目的是删除重复数据，且id是最小值的行不删除：

delete from jd_chapter a where a.`id` in
(select `id` from jd_chapter group by book_id,chapter_id having count(*)>1)
and a.`id` not in
(select min(`id`) from jd_chapter group by book_id,chapter_id having count(*)>1);

因为表大（千万级别），且使用了两个子查询，执行了很久没有执行完。
 

--------------------------思路----------------------------

采用临时表做关联，以下是步骤：

一、先到Slave库上把重复数据导出来，避免造成主库压力过大。

select id from jd_chapter group by book_id,chapter_id having count(*)>1 order by id asc
into outfile '/tmp/jd_chapter.sql' FIELDS TERMINATED BY ',';

二、拷贝导出的SQL到Master主库的/tmp/目录下

三、在Master主库上，建立一张临时表，并创建主键：

mysql> create TEMPORARY table tmp(id int,primary key(id));
Query OK, 0 rows affected (0.07 sec)

四、在Master主库上，LOAD方式导入至临时表里

load data infile '/tmp/jd_chapter.sql' into table tmp FIELDS TERMINATED BY ',';

五、在Master主库上，删除临时表最小的id

delete from tmp limit 1;

六、用临时表做关联，删除jd_chapter表重复数据

delete a from jd_chapter join tmp b on a.id=b.id;