MySQL中字符串查询效率大比拼
背景
最近有个同事对字符串加索引,加完后,发现多了个奇奇怪怪的数字
执行的SQL如下:
alter table string_index_test add index `idx_name` (`name`) USING BTREE;
这个奇怪数字就是191,它很是疑惑,也没指定索引的长度
通过查看MySQL官方文档
InnoDB has a maximum index length of 767 bytes for tables that use COMPACT or REDUNDANT row format, so for utf8mb3 or utf8mb4 columns, you can index a maximum of 255 or 191 characters, respectively. If you currently have utf8mb3 columns with indexes longer than 191 characters, you must index a smaller number of characters.
In an InnoDB table that uses COMPACT or REDUNDANT row format, these column and index definitions are legal:
col1 VARCHAR(500) CHARACTER SET utf8, INDEX (col1(255))
To use utf8mb4 instead, the index must be smaller:
col1 VARCHAR(500) CHARACTER SET utf8mb4, INDEX (col1(191))
大概意思就是InnoDB最大索引长度为 767 字节数,用的编码是utf8mb4,则可以存储191个字符(767/4 约等于 191),编码字段长度超出最大索引长度后MySQL 默认在普通索引追加了191
对字符串加索引?
一般情况下,是不建议在字符串加索引,占空间
如果一定要加,建议可以指定长度,前提是字符串前面部分区分度好的话,此时这类索引就叫前缀索引
区分度不好的话,很容易发生碰撞,进而引发一系列问题
我们再通过执行计划来分析一波
上面分别演示了前缀索引和普通索引在只有where条件、order by和group by不同执行情况,可以看到Extra的说明,前缀索引只有where条件,无法使用覆盖索引,order by会使用filesort,group by会使用temporary和filesort
总的来说,前缀索引无法使用覆盖索引,进而导致order by和group by要使用文件排序,甚至临时表前缀索引有这么些问题,不指定长度?怎么处理?
准备了单表100W的数据进行测试
使用性能压力测试工具mysqlslap
性能测试脚本
mysqlslap -uroot -p --concurrency=100,200 --iterations=1 --number-of-queries=1 --create-schema=test --query=C:\xxx\query.sql
–concurrency=100,200 测试并发的线程数/客户端数,第一次100,第二次200
–iterations=1 指定测试重复次数1次
–number-of-queries=1 指定每个线程执行的 SQL 语句数量上限(不精确)
–create-schema=test 指定查询的数据库test
1、不加索引
查询的SQL:SELECT SQL_NO_CACHE * FROM string_index_test WHERE name=‘forlan’;
Benchmark
Average number of seconds to run all queries: 8.328 seconds
Minimum number of seconds to run all queries: 8.328 seconds
Maximum number of seconds to run all queries: 8.328 seconds
Number of clients running queries: 100
Average number of queries per client: 0
Benchmark
Average number of seconds to run all queries: 18.078 seconds
Minimum number of seconds to run all queries: 18.078 seconds
Maximum number of seconds to run all queries: 18.078 seconds
Number of clients running queries: 200
Average number of queries per client: 0
2、加字符串索引
alter table string_index_test add index idx_name
(name
) USING BTREE;
查询的SQL:SELECT SQL_NO_CACHE * FROM string_index_test WHERE name=‘forlan’;
Benchmark
Average number of seconds to run all queries: 0.250 seconds
Minimum number of seconds to run all queries: 0.250 seconds
Maximum number of seconds to run all queries: 0.250 seconds
Number of clients running queries: 100
Average number of queries per client: 0
Benchmark
Average number of seconds to run all queries: 1.438 seconds
Minimum number of seconds to run all queries: 1.438 seconds
Maximum number of seconds to run all queries: 1.438 seconds
Number of clients running queries: 200
Average number of queries per client: 0
3、使用CRC32创建索引
CRC全称为Cyclic Redundancy Check,又叫循环冗余校验。
CRC32是CRC算法的一种,返回值的范围0~2^32-1,使用bigint存储
加一个name_crc32列,创建这个列的所有,索引空间小很多,利用整型加速查询
加索引:alter table string_index_test add index idx_nam_crc32
(name_crc32
) USING BTREE;
查询的SQL:SELECT SQL_NO_CACHE * FROM string_index_test WHERE name_crc32=CRC32(‘forlan’) and name=‘forlan’;因为CRC32存在发生碰撞,所以加上name条件,才能筛选出正确的数据
Benchmark
Average number of seconds to run all queries: 0.266 seconds
Minimum number of seconds to run all queries: 0.266 seconds
Maximum number of seconds to run all queries: 0.266 seconds
Number of clients running queries: 100
Average number of queries per client: 0
Benchmark
Average number of seconds to run all queries: 0.390 seconds
Minimum number of seconds to run all queries: 0.390 seconds
Maximum number of seconds to run all queries: 0.390 seconds
Number of clients running queries: 200
Average number of queries per client: 0
- 通过对字符串加索引,可以提高查询效率,但需要注意指定长度,无法使用覆盖索引
- 通过使用CRC32,需要额外存一个字段,将字符串转为整数存储,节省空间,效率提升并不是很大,但存在碰撞问题,可以加多字符串筛选条件
- -对于CRC32存在碰撞问题,可以使用CRC64减少碰撞,但需要安装 common_schema database函数库