时间:2021-12-01 10:31:38 | 栏目:Mysql | 点击:次
UNION语义:取两个子查询结果的并集,重复的行只保留一行
CREATE TABLE t1(id INT PRIMARY KEY, a INT, b INT, INDEX(a)); DELIMITER ;; CREATE PROCEDURE idata() BEGIN DECLARE i INT; SET i=1; WHILE (i<= 1000) DO INSERT INTO t1 VALUES (i,i,i); SET i=i+1; END WHILE; END;; DELIMITER ; CALL idata();
(SELECT 1000 AS f) UNION (SELECT id FROM t1 ORDER BY id DESC LIMIT 2); mysql> EXPLAIN (SELECT 1000 AS f) UNION (SELECT id FROM t1 ORDER BY id DESC LIMIT 2); +----+--------------+------------+------------+-------+---------------+---------+---------+------+------+----------+----------------------------------+ | id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra | +----+--------------+------------+------------+-------+---------------+---------+---------+------+------+----------+----------------------------------+ | 1 | PRIMARY | NULL | NULL | NULL | NULL | NULL | NULL | NULL | NULL | NULL | No tables used | | 2 | UNION | t1 | NULL | index | NULL | PRIMARY | 4 | NULL | 2 | 100.00 | Backward index scan; Using index | | NULL | UNION RESULT | <union1,2> | NULL | ALL | NULL | NULL | NULL | NULL | NULL | NULL | Using temporary | +----+--------------+------------+------------+-------+---------------+---------+---------+------+------+----------+----------------------------------+
第二行的Key=PRIMARY
,Using temporary
UNION RESULT
的时候,使用了临时表UNION ALL
没有去重的语义,一次执行子查询,得到的结果直接发给客户端,不需要内存临时表
mysql> EXPLAIN (SELECT 1000 AS f) UNION ALL (SELECT id FROM t1 ORDER BY id DESC LIMIT 2); +----+-------------+-------+------------+-------+---------------+---------+---------+------+------+----------+----------------------------------+ | id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra | +----+-------------+-------+------------+-------+---------------+---------+---------+------+------+----------+----------------------------------+ | 1 | PRIMARY | NULL | NULL | NULL | NULL | NULL | NULL | NULL | NULL | NULL | No tables used | | 2 | UNION | t1 | NULL | index | NULL | PRIMARY | 4 | NULL | 2 | 100.00 | Backward index scan; Using index | +----+-------------+-------+------------+-------+---------------+---------+---------+------+------+----------+----------------------------------+
-- 16777216 Bytes = 16 MB mysql> SHOW VARIABLES like '%tmp_table_size%'; +----------------+----------+ | Variable_name | Value | +----------------+----------+ | tmp_table_size | 16777216 | +----------------+----------+
-- MySQL 5.6上执行 mysql> EXPLAIN SELECT id%10 AS m, COUNT(*) AS c FROM t1 GROUP BY m; +----+-------------+-------+-------+---------------+------+---------+------+------+----------------------------------------------+ | id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra | +----+-------------+-------+-------+---------------+------+---------+------+------+----------------------------------------------+ | 1 | SIMPLE | t1 | index | PRIMARY,a | a | 5 | NULL | 1000 | Using index; Using temporary; Using filesort | +----+-------------+-------+-------+---------------+------+---------+------+------+----------------------------------------------+ mysql> SELECT id%10 AS m, COUNT(*) AS c FROM t1 GROUP BY m; +------+-----+ | m | c | +------+-----+ | 0 | 100 | | 1 | 100 | | 2 | 100 | | 3 | 100 | | 4 | 100 | | 5 | 100 | | 6 | 100 | | 7 | 100 | | 8 | 100 | | 9 | 100 | +------+-----+
Using index
:表示使用了覆盖索引,选择了索引a,不需要回表
Using temporary
:表示使用了临时表
Using filesort
:表示需要排序
(x,1)
-- 跳过最后的排序阶段,直接从临时表中取回数据 mysql> EXPLAIN SELECT id%10 AS m, COUNT(*) AS c FROM t1 GROUP BY m ORDER BY NULL; +----+-------------+-------+-------+---------------+------+---------+------+------+------------------------------+ | id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra | +----+-------------+-------+-------+---------------+------+---------+------+------+------------------------------+ | 1 | SIMPLE | t1 | index | PRIMARY,a | a | 5 | NULL | 1000 | Using index; Using temporary | +----+-------------+-------+-------+---------------+------+---------+------+------+------------------------------+ -- t1中的数据是从1开始的 mysql> SELECT id%10 AS m, COUNT(*) AS c FROM t1 GROUP BY m ORDER BY NULL; +------+-----+ | m | c | +------+-----+ | 1 | 100 | | 2 | 100 | | 3 | 100 | | 4 | 100 | | 5 | 100 | | 6 | 100 | | 7 | 100 | | 8 | 100 | | 9 | 100 | | 0 | 100 | +------+-----+
SET tmp_table_size=1024;
-- 内存临时表的上限为1024 Bytes,但内存临时表不能完全放下100行数据,内存临时表会转成磁盘临时表,默认采用InnoDB引擎 -- 如果t1很大,这个查询需要的磁盘临时表就会占用大量的磁盘空间 mysql> SELECT id%100 AS m, count(*) AS c FROM t1 GROUP BY m ORDER BY NULL LIMIT 10; +------+----+ | m | c | +------+----+ | 1 | 10 | | 2 | 10 | | 3 | 10 | | 4 | 10 | | 5 | 10 | | 6 | 10 | | 7 | 10 | | 8 | 10 | | 9 | 10 | | 10 | 10 | +------+----+
不论使用内存临时表还是磁盘临时表,GROUP BY
都需要构造一个带唯一索引的表,执行代价较高
需要临时表的原因:每一行的id%100
是无序的,因此需要临时表,来记录并统计结果
如果可以确保输入的数据是有序的,那么计算GROUP BY
时,只需要
从左到右顺序扫描,依次累加即可
(0,X)
(1,Y)
-- MySQL 5.7上执行 ALTER TABLE t1 ADD COLUMN z INT GENERATED ALWAYS AS(id % 100), ADD INDEX(z); -- 使用了覆盖索引,不需要临时表,也不需要排序 mysql> EXPLAIN SELECT z, COUNT(*) AS c FROM t1 GROUP BY z; +----+-------------+-------+------------+-------+---------------+------+---------+------+------+----------+-------------+ | id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra | +----+-------------+-------+------------+-------+---------------+------+---------+------+------+----------+-------------+ | 1 | SIMPLE | t1 | NULL | index | z | z | 5 | NULL | 1000 | 100.00 | Using index | +----+-------------+-------+------------+-------+---------------+------+---------+------+------+----------+-------------+ 2
一个GROUP BY
语句需要放到临时表的数据量特别大,还是按照先放在内存临时表,再退化成磁盘临时表
可以直接用磁盘临时表的形式,在GROUP BY
语句中SQL_BIG_RESULT
(告诉优化器涉及的数据量很大)
磁盘临时表原本采用B+树存储,存储效率还不如数组,优化器看到SQL_BIG_RESULT
,会直接用数组存储
-- 没有再使用临时表,而是直接使用了排序算法 mysql> EXPLAIN SELECT SQL_BIG_RESULT id%100 AS m, COUNT(*) AS c FROM t1 GROUP BY m; +----+-------------+-------+-------+---------------+------+---------+------+------+-----------------------------+ | id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra | +----+-------------+-------+-------+---------------+------+---------+------+------+-----------------------------+ | 1 | SIMPLE | t1 | index | PRIMARY,a | a | 5 | NULL | 1000 | Using index; Using filesort | +----+-------------+-------+-------+---------------+------+---------+------+------+-----------------------------+
初始化sort_buffer
,确定放入一个整型字段,记为m
扫描t1的索引a,依次取出里面的id值,将id%100的值放入sort_buffer
扫描完成后,对sort_buffer
的字段m做排序(sort_buffer内存不够时,会利用磁盘临时文件辅助排序)
排序完成后,得到一个有序数组,遍历有序数组,得到每个值出现的次数(类似上面优化索引的方式)
-- 标准SQL,SELECT部分添加一个聚合函数COUNT(*) SELECT a,COUNT(*) FROM t GROUP BY a ORDER BY NULL; -- 非标准SQL SELECT a FROM t GROUP BY a ORDER BY NULL; SELECT DISTINCT a FROM t;
标准SQL:按照字段a分组,计算每组a出现的次数
非标准SQL:没有了COUNT(*)
,不再需要执行计算总数的逻辑
DISTINCT
语义一致如果不需要执行聚合函数 ,DISTINCT
和GROUP BY
的语义、执行流程和执行性能是相同的
join_buffer
是无序数组,sort_buffer
是有序数组,临时表是二维表结构GROUP BY
语句的结果没有明确的排序要求,加上ORDER BY NULL
(MySQL 5.6)GROUP BY
过程用上索引,确认EXPLAIN结果没有Using temporary
和Using filesort
GROUP BY
需要统计的数据量不大,尽量使用内存临时表(可以适当调大tmp_table_size
)SQL_BIG_RESULT
来告诉优化器直接使用排序算法(跳过临时表)