删除一个表中的重复数据同时保留第一次插入那一条以及sql优化

       业务：一个表中有很多数据（id为自增主键），在这些数据中有个别数据出现了重复的数据。

       目标：需要把这些重复数据删除同时保留第一次插入的那一条数据，还要保持其它的数据不受影响。

       解题过程：

       第一步：查出所有要保留的下来的数据的id（save_id）

SELECT id as save_id
FROM yujing.alarm_event_info_snapshot aeis
where aeis.event_id in
(SELECT ae.id
FROM yujing.alarm_event ae
where ae.event_uuid like 'yuanwtj_%')
group by (aeis.event_id)

优化后：

SELECT aeis.id as save_id
FROM yujing.alarm_event ae
right join yujing.alarm_event_info_snapshot aeis
on aeis.event_id = ae.id
where ae.event_uuid like 'yuanwtj_%'
group by (aeis.event_id)

  

       第二步：获取所有相关数据的id（all_id）

SELECT aeis.id as all_id
FROM yujing.alarm_event_info_snapshot aeis
where aeis.event_id in
(SELECT ae.id
FROM yujing.alarm_event ae
where ae.event_uuid like 'yuanwtj_%')
order by aeis.event_id

优化后：       

SELECT aeis.id as all_id
FROM yujing.alarm_event ae
right join yujing.alarm_event_info_snapshot aeis
on aeis.event_id = ae.id
where ae.event_uuid like 'yuanwtj_%'

        第三步：获取要删除的数据的id（del_id）

select ad.all_id as del_id
from (SELECT aeis.id as all_id
FROM yujing.alarm_event_info_snapshot aeis
where aeis.event_id in
(SELECT ae.id
FROM yujing.alarm_event ae
where ae.event_uuid like 'yuanwtj_%')) as ad
where ad.all_id not in (SELECT id as save_id
FROM yujing.alarm_event_info_snapshot aeis
where aeis.event_id in
(SELECT ae.id
FROM yujing.alarm_event ae
where ae.event_uuid like 'yuanwtj_%')
group by (aeis.event_id))

优化后：

select ad.all_id as del_id
from (SELECT aeis.id as all_id
FROM yujing.alarm_event ae
right join yujing.alarm_event_info_snapshot aeis
on aeis.event_id = ae.id
where ae.event_uuid like 'yuanwtj_%') as ad
left join (SELECT aeis.id as save_id
FROM yujing.alarm_event ae
right join yujing.alarm_event_info_snapshot aeis
on aeis.event_id = ae.id
where ae.event_uuid like 'yuanwtj_%'
group by (aeis.event_id)) as sd
on ad.all_id = sd.save_id
where sd.save_id is null

       

       第四步：根据id删除所有节点，注意mysql中如果有大量数据时需要批量删除，我最后使用了ETL工具进行的批量删除
       总结：在mysql数据库中，sql语句中最好不要在in或not in关键字的查询里动态获取匹配的值，数据量大的情况下使用它们效率很低，可以使用左右连接来代替in操作，这样效率会提高很多倍，大数据量下尤为明显。