整个表target_table大概有14000W条记录，有个主键id,需要去重的字段是mac.

1. 查看重复的记录数

   SELECT
    min( id ) AS mid,
    mac,
    count ( * ) 
   FROM
    target_table 
   GROUP BY
    mac 
   HAVING
    count( mac ) > 1;
   

   整个表重复的数据2w+

2. 创建临时表

   CREATE TABLE dup_tmp (
   SELECT
    min( id ) AS mid,
    mac,
    count ( * ) 
   FROM
    target_table 
   GROUP BY
    mac 
   HAVING
    count( mac ) > 1 
    );
   

   将重复的mac筛选到临时表

3. 删除多余的数据保留最小的id记录

   DELETE target 
   FROM
    target_table AS target
    JOIN dup_tmp AS tmp ON target.mac = tmp.mac 
   WHERE
    target.id > tmp.mid;
   

   刚开始的时候mac字段没有建索引，删除很慢。建了索引删除就很快了。