Mysql 由单引号引发的性能问题

最新推荐文章于 2024-08-23 22:00:00 发布

_lijunru

最新推荐文章于 2024-08-23 22:00:00 发布

阅读量333

点赞数

分类专栏： MySQL 文章标签： MySQL

本文链接：https://blog.csdn.net/weixin_39350124/article/details/91368616

版权

MySQL 专栏收录该内容

0 篇文章 0 订阅

订阅专栏

最近工作的时候遇到个问题就是要去优化一条Mysql的查询，想了许多方法都不尽人意，没能把查询速度提高到理想的速度，后来无意间把查询的条件加上单引号，性能就大大提高了，为什么加了单引号性能就会变得这么快呢？
优化前的sql

select  table_name,yet_totalSize,day1_totalSize,day1_rate,day7_rate,day7_totalSize from (select r.tbl_name as table_name,sum(if(r.d_s in (20190602) ,r.totalsize,0 ) ) as yet_totalSize,
                sum( if(r.d_s in (20190601) ,r.totalsize,0) ) as day1_totalSize,
                (sum( if(r.d_s in (20190602) ,r.totalsize,0) ) - sum(if(r.d_s in (20190601) ,r.totalsize,0 )))/sum(if(r.d_s in (20190601) ,r.totalsize,0 ) ) as day1_rate,
                (sum( if(r.d_s in (20190602) ,r.totalsize,0) ) - sum(if(r.d_s in (20190526) ,r.totalsize,0 )))/sum(if(r.d_s in (20190526) ,r.totalsize,0 ) ) as day7_rate,
                sum( if(r.d_s in (20190526) ,r.totalsize,0) ) as day7_totalSize
                from uts3.monitor_data_rawlog r where r.tbl_name  in ( "...") and   r.d_s in (20190602,20190601,20190526) group by  r.tbl_name) t order by day1_totalSize desc limit 0,10

优化后的sql

select  table_name,yet_totalSize,day1_totalSize,day1_rate,day7_rate,day7_totalSize from (select r.tbl_name as table_name,sum(if(r.d_s in ('20190602') ,r.totalsize,0 ) ) as yet_totalSize,
                sum( if(r.d_s in ('20190601') ,r.totalsize,0) ) as day1_totalSize,
                (sum( if(r.d_s in ('20190602') ,r.totalsize,0) ) - sum(if(r.d_s in ('20190601') ,r.totalsize,0 )))/sum(if(r.d_s in ('20190601') ,r.totalsize,0 ) ) as day1_rate,
                (sum( if(r.d_s in ('20190602') ,r.totalsize,0) ) - sum(if(r.d_s in ('20190526') ,r.totalsize,0 )))/sum(if(r.d_s in ('20190526') ,r.totalsize,0 ) ) as day7_rate,
                sum( if(r.d_s in ('20190526') ,r.totalsize,0) ) as day7_totalSize
                from uts3.monitor_data_rawlog r where r.tbl_name  in ( "...") and   r.d_s in ('20190602','20190601','20190526') group by  r.tbl_name) t order by day1_totalSize desc limit 0,10

注意只是d_s这个字段补上单引号而已。
先用EXPLAIN分析优化前的sql语句：
在这里插入图片描述
再分析优化后的sql语句：
从2.67s降到170ms，可以看出优化前type=index，key=idx_monitor_data_rawlog_tn（table_name的索引），表示查询是读全表的，没有使用到d_s的索引，扫描了780675行数据。
再看优化后的sql，type=range，key=inx_ds_tn（d_s和table_name的聚合索引），表示查询是按索引范围扫描的，d_s的索引有被使用到，总共才扫描20335行数据。
所以可以看出，没加引号之前，d_s的索引是没有被用到的。
为什么呢？
因为d_s是varchar类型的，而没加单引号字段会被当作数字类型来使用，索引就没法正常使用了（索引）。这就是索引失效的一个条件之一：varchar类型在使用索引的时候要用单引号。
总结：如果列类型是字符串，那一定要在条件中将数据使用单引号包含起来，否则索引会失效。