性能优化
MySQL在2016年仍然保持强劲的数据库流行度增长趋势。越来越多的客户将自己的应用建立在MySQL数据库之上,甚至是从Oracle迁移到MySQL上来。但也存在部分客户在使用MySQL数据库的过程中遇到一些比如响应时间慢,CPU打满等情况。阿里云RDS专家服务团队帮助云上客户解决过很多紧急问题。现将《ApsaraDB专家诊断报告》中出现的部分常见SQL问题总结如下,供大家参考。
常见SQL错误用法
1. LIMIT 语句
分页查询是最常用的场景之一,但也通常也是最容易出问题的地方。比如对于下面简单的语句,一般DBA想到的办法是在type, name, create_time字段上加组合索引。这样条件排序都能有效的利用到索引,性能迅速提升。
SELECT * FROM operation WHERE type = 'SQLStats' AND name = 'SlowLog' ORDER BY create_time LIMIT 1000, 10;
好吧,可能90%以上的DBA解决该问题就到此为止。但当 LIMIT 子句变成 “LIMIT 1000000,10” 时,程序员仍然会抱怨:我只取10条记录为什么还是慢?
要知道数据库也并不知道第1000000条记录从什么地方开始,即使有索引也需要从头计算一次。出现这种性能问题,多数情形下是程序员偷懒了。在前端数据浏览翻页,或者大数据分批导出等场景下,是可以将上一页的最大值当成参数作为查询条件的。SQL重新设计如下:
SELECT * FROM operation WHERE type = 'SQLStats' AND name = 'SlowLog' AND create_time > '2017-03-16 14:00:00' ORDER BY create_time limit 10;
在新设计下查询时间基本固定,不会随着数据量的增长而发生变化。
2. 隐式转换
SQL语句中查询变量和字段定义类型不匹配是另一个常见的错误。比如下面的语句:
mysql> explain extended SELECT * > FROM my_balance b > WHERE b.bpn = 14000000123 > AND b.isverified IS NULL ; mysql> show warnings; | Warning | 1739 | Cannot use ref access on index 'bpn' due to type or collation conversion on field 'bpn'
其中字段bpn的定义为varchar(20),MySQL的策略是将字符串转换为数字之后再比较。函数作用于表字段,索引失效。
上述情况可能是应用程序框架自动填入的参数,而不是程序员的原意。现在应用框架很多很繁杂,使用方便的同时也小心它可能给自己挖坑。
3. 关联更新、删除
虽然MySQL5.6引入了物化特性,但需要特别注意它目前仅仅针对查询语句的优化。对于更新或删除需要手工重写成JOIN。
比如下面UPDATE语句,MySQL实际执行的是循环/嵌套子查询(DEPENDENT SUBQUERY),其执行时间可想而知。
UPDATE operation o SET status = 'applying' WHERE o.id IN (SELECT id FROM (SELECT o.id, o.status FROM operation o WHERE o.group = 123 AND o.status NOT IN ( 'done' ) ORDER BY o.parent, o.id LIMIT 1) t);
执行计划:
+----+--------------------+-------+-------+---------------+---------+---------+-------+------+-----------------------------------------------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+--------------------+-------+-------+---------------+---------+---------+-------+------+-----------------------------------------------------+
| 1 | PRIMARY | o | index | | PRIMARY | 8 | | 24 | Using where; Using temporary |
| 2 | DEPENDENT SUBQUERY | | | | | | | | Impossible WHERE noticed after reading const tables |
| 3 | DERIVED | o | ref | idx_2,idx_5 | idx_5 | 8 | const | 1 | Using where; Using filesort |
+----+--------------------+-------+-------+---------------+---------+---------+-------+------+-----------------------------------------------------+
重写为JOIN之后,子查询的选择模式从DEPENDENT SUBQUERY变成DERIVED,执行速度大大加快,从7秒降低到2毫秒。
UPDATE operation o JOIN (SELECT o.id, o.status FROM operation o WHERE o.group = 123 AND o.status NOT IN ( 'done' ) ORDER BY o.parent, o.id LIMIT 1) t ON o.id = t.id SET status = 'applying'
执行计划简化为:
+----+-------------+-------+------+---------------+-------+---------+-------+------+-----------------------------------------------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+-------------+-------+------+---------------+-------+---------+-------+------+-----------------------------------------------------+
| 1 | PRIMARY | | | | | | | | Impossible WHERE noticed after reading const tables |
| 2 | DERIVED | o | ref | idx_2,idx_5 | idx_5 | 8 | const | 1 | Using where; Using filesort |
+----+-------------+-------+------+---------------+-------+---------+-------+------+-----------------------------------------------------+
4. 混合排序
MySQL不能利用索引进行混合排序。但在某些场景,还是有机会使用特殊方法提升性能的。
SELECT * FROM my_order o INNER JOIN my_appraise a ON a.orderid = o.id ORDER BY a.is_reply ASC, a.appraise_time DESC LIMIT 0, 20
执行计划显示为全表扫描:
+----+-------------+-------+--------+-------------+---------+---------+---------------+---------+-+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra
+----+-------------+-------+--------+-------------+---------+---------+---------------+---------+-+
| 1 | SIMPLE | a | ALL | idx_orderid | NULL | NULL | NULL | 1967647 | Using filesort |
| 1 | SIMPLE | o | eq_ref | PRIMARY | PRIMARY | 122 | a.orderid | 1 | NULL |
+----+-------------+-------+--------+---------+---------+---------+-----------------+---------+-+
由于is_reply只有0和1两种状态,我们按照下面的方法重写后,执行时间从1.58秒降低到2毫秒。
SELECT * FROM ((SELECT * FROM my_order o INNER JOIN my_appraise a ON a.orderid = o.id AND is_reply = 0 ORDER BY appraise_time DESC LIMIT 0, 20) UNION ALL (SELECT * FROM my_order o INNER JOIN my_appraise a ON a.orderid = o.id AND is_reply = 1 ORDER BY appraise_time DESC LIMIT 0, 20)) t ORDER BY is_reply ASC, appraisetime DESC LIMIT 20;
5. EXISTS语句
MySQL对待EXISTS子句时,仍然采用嵌套子查询的执行方式。如下面的SQL语句:
SELECT * FROM my_neighbor n LEFT JOIN my_neighbor_apply sra ON n.id = sra.neighbor_id AND sra.user_id = 'xxx' WHERE n.topic_status < 4 AND EXISTS(SELECT 1 FROM message_info m WHERE n.id = m.neighbor_id AND m.inuser = 'xxx') AND n.topic_type <> 5
执行计划为:
+----+--------------------+-------+------+-----+------------------------------------------+---------+-------+---------+ -----+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+--------------------+-------+------+ -----+------------------------------------------+---------+-------+---------+ -----+
| 1 | PRIMARY | n | ALL | | NULL | NULL | NULL | 1086041 | Using where |
| 1 | PRIMARY | sra | ref | | idx_user_id | 123 | const | 1 | Using where |
| 2 | DEPENDENT SUBQUERY | m | ref | | idx_message_info | 122 | const | 1 | Using index condition; Using where |
+----+--------------------+-------+------+ -----+------------------------------------------+---------+-------+---------+ -----+
去掉exists更改为join,能够避免嵌套子查询,将执行时间从1.93秒降低为1毫秒。
SELECT * FROM my_neighbor n INNER JOIN message_info m ON n.id = m.neighbor_id AND m.inuser = 'xxx' LEFT JOIN my_neighbor_apply sra ON n.id = sra.neighbor_id AND sra.user_id = 'xxx' WHERE n.topic_status < 4 AND n.topic_type <> 5
新的执行计划:
+----+-------------+-------+--------+ -----+------------------------------------------+---------+ -----+------+ -----+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+-------------+-------+--------+ -----+------------------------------------------+---------+ -----+------+ -----+
| 1 | SIMPLE | m | ref | | idx_message_info | 122 | const | 1 | Using index condition |
| 1 | SIMPLE | n | eq_ref | | PRIMARY | 122 | ighbor_id | 1 | Using where |
| 1 | SIMPLE | sra | ref | | idx_user_id | 123 | const | 1 | Using where |
+----+-------------+-------+--------+ -----+------------------------------------------+---------+ -----+------+ -----+
6. 条件下推
外部查询条件不能够下推到复杂的视图或子查询的情况有:
- 聚合子查询;
- 含有LIMIT的子查询;
- UNION 或UNION ALL子查询;
- 输出字段中的子查询;
如下面的语句,从执行计划可以看出其条件作用于聚合子查询之后:
SELECT * FROM (SELECT target, Count(*) FROM operation GROUP BY target) t WHERE target = 'rm-xxxx'
+----+-------------+------------+-------+---------------+-------------+---------+-------+------+-------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+-------------+------------+-------+---------------+-------------+---------+-------+------+-------------+
| 1 | PRIMARY | <derived2> | ref | <auto_key0> | <auto_key0> | 514 | const | 2 | Using where |
| 2 | DERIVED | operation | index | idx_4 | idx_4 | 519 | NULL | 20 | Using index |
+----+-------------+------------+-------+---------------+-------------+---------+-------+------+-------------+
确定从语义上查询条件可以直接下推后,重写如下:
SELECT target, Count(*) FROM operation WHERE target = 'rm-xxxx' GROUP BY target
执行计划变为:
+----+-------------+-----------+------+---------------+-------+---------+-------+------+--------------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+-------------+-----------+------+---------------+-------+---------+-------+------+--------------------+
| 1 | SIMPLE | operation | ref | idx_4 | idx_4 | 514 | const | 1 | Using where; Using index |
+----+-------------+-----------+------+---------------+-------+---------+-------+------+--------------------+
关于MySQL外部条件不能下推的详细解释说明请参考以前文章:MySQL · 性能优化 · 条件下推到物化表
7. 提前缩小范围
先上初始SQL语句:
SELECT * FROM my_order o LEFT JOIN my_userinfo u ON o.uid = u.uid LEFT JOIN my_productinfo p ON o.pid = p.pid WHERE ( o.display = 0 ) AND ( o.ostaus = 1 ) ORDER BY o.selltime DESC LIMIT 0, 15
该SQL语句原意是:先做一系列的左连接,然后排序取前15条记录。从执行计划也可以看出,最后一步估算排序记录数为90万,时间消耗为12秒。
+----+-------------+-------+--------+---------------+---------+---------+-----------------+--------+----------------------------------------------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+-------------+-------+--------+---------------+---------+---------+-----------------+--------+----------------------------------------------------+
| 1 | SIMPLE | o | ALL | NULL | NULL | NULL | NULL | 909119 | Using where; Using temporary; Using filesort |
| 1 | SIMPLE | u | eq_ref | PRIMARY | PRIMARY | 4 | o.uid | 1 | NULL |
| 1 | SIMPLE | p | ALL | PRIMARY | NULL | NULL | NULL | 6 | Using where; Using join buffer (Block Nested Loop) |
+----+-------------+-------+--------+---------------+---------+---------+-----------------+--------+----------------------------------------------------+
由于最后WHERE条件以及排序均针对最左主表,因此可以先对my_order排序提前缩小数据量再做左连接。SQL重写后如下,执行时间缩小为1毫秒左右。
SELECT * FROM ( SELECT * FROM my_order o WHERE ( o.display = 0 ) AND ( o.ostaus = 1 ) ORDER BY o.selltime DESC LIMIT 0, 15 ) o LEFT JOIN my_userinfo u ON o.uid = u.uid LEFT JOIN my_productinfo p ON o.pid = p.pid ORDER BY o.selltime DESC limit 0, 15
再检查执行计划:子查询物化后(select_type=DERIVED)参与JOIN。虽然估算行扫描仍然为90万,但是利用了索引以及LIMIT 子句后,实际执行时间变得很小。
+----+-------------+------------+--------+---------------+---------+---------+-------+--------+----------------------------------------------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+-------------+------------+--------+---------------+---------+---------+-------+--------+----------------------------------------------------+
| 1 | PRIMARY | <derived2> | ALL | NULL | NULL | NULL | NULL | 15 | Using temporary; Using filesort |
| 1 | PRIMARY | u | eq_ref | PRIMARY | PRIMARY | 4 | o.uid | 1 | NULL |
| 1 | PRIMARY | p | ALL | PRIMARY | NULL | NULL | NULL | 6 | Using where; Using join buffer (Block Nested Loop) |
| 2 | DERIVED | o | index | NULL | idx_1 | 5 | NULL | 909112 | Using where |
+----+-------------+------------+--------+---------------+---------+---------+-------+--------+----------------------------------------------------+
8. 中间结果集下推
再来看下面这个已经初步优化过的例子(左连接中的主表优先作用查询条件):
SELECT a.*, c.allocated FROM ( SELECT resourceid FROM my_distribute d WHERE isdelete = 0 AND cusmanagercode = '1234567' ORDER BY salecode limit 20) a LEFT JOIN ( SELECT resourcesid, sum(ifnull(allocation, 0) * 12345) allocated FROM my_resources GROUP BY resourcesid) c ON a.resourceid = c.resourcesid
那么该语句还存在其它问题吗?不难看出子查询 c 是全表聚合查询,在表数量特别大的情况下会导致整个语句的性能下降。
其实对于子查询 c,左连接最后结果集只关心能和主表resourceid能匹配的数据。因此我们可以重写语句如下,执行时间从原来的2秒下降到2毫秒。
SELECT a.*, c.allocated FROM ( SELECT resourceid FROM my_distribute d WHERE isdelete = 0 AND cusmanagercode = '1234567' ORDER BY salecode limit 20) a LEFT JOIN ( SELECT resourcesid, sum(ifnull(allocation, 0) * 12345) allocated FROM my_resources r, ( SELECT resourceid FROM my_distribute d WHERE isdelete = 0 AND cusmanagercode = '1234567' ORDER BY salecode limit 20) a WHERE r.resourcesid = a.resourcesid GROUP BY resourcesid) c ON a.resourceid = c.resourcesid
但是子查询 a 在我们的SQL语句中出现了多次。这种写法不仅存在额外的开销,还使得整个语句显的繁杂。使用WITH语句再次重写:
WITH a AS ( SELECT resourceid FROM my_distribute d WHERE isdelete = 0 AND cusmanagercode = '1234567' ORDER BY salecode limit 20) SELECT a.*, c.allocated FROM a LEFT JOIN ( SELECT resourcesid, sum(ifnull(allocation, 0) * 12345) allocated FROM my_resources r, a WHERE r.resourcesid = a.resourcesid GROUP BY resourcesid) c ON a.resourceid = c.resourcesid
AliSQL即将推出WITH语法,敬请期待。
▌多线程插入(单表)
问:为何对同一个表的插入多线程会比单线程快?同一时间对一个表的写操作不应该是独占的吗?
答:在数据里做插入操作的时候,整体时间的分配是这样的:
-
链接耗时 (30%)
-
发送 query 到服务器 (20%)
-
解析 query (20%)
-
插入操作 (10% * 词条数目)
-
插入 index (10% * Index的数目)
-
关闭链接 (10%)
从这里可以看出来,真正耗时的不是操作,而是链接,解析的过程。
MySQL 插入数据在写阶段是独占的,但是插入一条数据仍然需要解析、计算、最后才进行写处理,比如要给每一条记录分配自增 id,校验主键唯一键属性,或者其他一些逻辑处理,都是需要计算的,所以说多线程能够提高效率。
▌多线程插入(多表)
分区分表后使用多线程插入。
▌预处理 SQL
-
普通 SQL,即使用 Statement 接口执行 SQL
-
预处理 SQL,即使用 PreparedStatement 接口执行 SQL
使用 PreparedStatement 接口允许数据库预编译 SQL 语句,以后只需传入参数,避免了数据库每次都编译 SQL 语句,因此性能更好。
String sql = "insert into testdb.tuser (name, remark, createtime, updatetime) values (?, ?, ?, ?)"; for (int i = 0; i < m; i++) { //从池中获取连接 Connection conn = myBroker.getConnection(); PreparedStatement pstmt = conn.prepareStatement(sql); for (int k = 0; k < n; k++) { pstmt.setString(1, RandomToolkit.generateString(12)); pstmt.setString(2, RandomToolkit.generateString(24)); pstmt.setDate(3, new Date(System.currentTimeMillis())); pstmt.setDate(4, new Date(System.currentTimeMillis())); //加入批处理 pstmt.addBatch(); } pstmt.executeBatch(); //执行批处理 pstmt.close(); myBroker.freeConnection(conn); //连接归池 }复制代码
▌多值插入 SQL
-
普通插入 SQL:INSERT INTO TBL_TEST (id) VALUES(1)
-
多值插入 SQL:INSERT INTO TBL_TEST (id) VALUES (1), (2), (3)
使用多值插入 SQL,SQL 语句的总长度减少,即减少了网络 IO,同时也降低了连接次数,数据库一次 SQL 解析,能够插入多条数据。
▌事务( N 条提交一次)
在一个事务中提交大量 INSERT 语句可以提高性能。
1、将表的存储引擎修改为 myisam
2、将 sql 拼接成字符串,每 1000 条左右提交事务。
/// <summary> /// 执行多条SQL语句,实现数据库事务。 /// </summary>mysql数据库 /// <param name="SQLStringList">多条SQL语句</param> public void ExecuteSqlTran(List<string> SQLStringList) { using (MySqlConnection conn = new MySqlConnection(connectionString)) { if (DBVariable.flag) { conn.Open(); MySqlCommand cmd = new MySqlCommand(); cmd.Connection = conn; MySqlTransaction tx = conn.BeginTransaction(); cmd.Transaction = tx; try { for (int n = 0; n < SQLStringList.Count; n++) { string strsql = SQLStringList[n].ToString(); if (strsql.Trim().Length > 1) { cmd.CommandText = strsql; cmd.ExecuteNonQuery(); } //后来加上的 if (n > 0 && (n % 1000 == 0 || n == SQLStringList.Count - 1)) { tx.Commit(); tx = conn.BeginTransaction(); } } //tx.Commit();//原来一次性提交 } catch (System.Data.SqlClient.SqlException E) { tx.Rollback(); throw new Exception(E.Message); } } } }
10w 条数据大概用时 10s!
总结
- 数据库编译器产生执行计划,决定着SQL的实际执行方式。但是编译器只是尽力服务,所有数据库的编译器都不是尽善尽美的。上述提到的多数场景,在其它数据库中也存在性能问题。了解数据库编译器的特性,才能避规其短处,写出高性能的SQL语句。
- 程序员在设计数据模型以及编写SQL语句时,要把算法的思想或意识带进来。
- 编写复杂SQL语句要养成使用WITH语句的习惯。简洁且思路清晰的SQL语句也能减小数据库的负担 ^^。
- 使用云上数据库遇到难点(不局限于SQL问题),随时寻求阿里云原厂专家服务的帮助。