HANDLER命令与实现

时间:2023-03-08 16:54:21
HANDLER命令与实现

MySQL“自古以来”都有一个神秘的HANDLER命令,而此命令非SQL标准语法,可以降低优化器对于SQL语句的解析与优化开销,从而提升查询性能。看到这里,可能有小伙伴不淡定了,这么好的东西为啥没广泛使用呢?这不是与几年前很夯的handlersocket插件类似吗?

那么,我们先来看看Handler语法说明:

HANDLER tbl_name OPEN [ [AS] alias]
HANDLER tbl_name READ index_name { = | <= | >= | < | > } (value1,value2,…) [ WHERE where_condition ] [LIMIT … ]
HANDLER tbl_name READ index_name { FIRST | NEXT | PREV | LAST } [ WHERE where_condition ] [LIMIT … ]
HANDLER tbl_name READ { FIRST | NEXT } [ WHERE where_condition ] [LIMIT … ]      //主健,无索引指定
HANDLER tbl_name CLOSE

首先从语法上看,HANDLER可以通过指定的索引去访问数据。但此语法并不支持DML操作。此外,由于减少了SQL解析,Handler命令的性能真的非常不错,根据Inside君的简单主键测试,Handler命令比SQL要快40%~45%。测试脚本如下:

SET @id=FLOOR(RAND()*1000000);
HANDLER sbtest.sbtest1 OPEN AS c;
HANDLER C READ `PRIMARY` = (@id);
HANDLER C CLOSE;

mysql> select * from test.test ;
+----+------+---------------------+
| id | data | ts |
+----+------+---------------------+
| 1 | abc | 2016-07-18 23:44:05 |
| 2 | abc | 2016-07-18 23:44:05 |
| 3 | abd | 2016-07-18 23:44:05 |
| 4 | acd | 2016-07-18 23:44:05 |
| 5 | def | 2016-07-18 23:44:05 |
| 6 | pqr | 2016-07-18 23:44:05 |
| 7 | stu | 2016-07-18 23:44:05 |
| 8 | vwx | 2016-07-18 23:44:05 |
| 9 | yza | 2016-07-18 23:44:05 |
| 10 | def | 2016-07-18 23:44:05 |
+----+------+---------------------+
10 rows in set (0.01 sec)
CREATE TABLE `test` (
`id` int(10) unsigned NOT NULL AUTO_INCREMENT,
`data` varchar(32) DEFAULT NULL,
`ts` timestamp NULL DEFAULT NULL,
PRIMARY KEY (`id`),
KEY `data` (`data`)
) ENGINE=InnoDB AUTO_INCREMENT=11 DEFAULT CHARSET=gbk
mysql> handler test.test open as c;    //打开
Query OK, 0 rows affected (0.01 sec) mysql> handler c read `PRIMARY`=(5); //查询主健
+----+------+---------------------+
| id | data | ts |
+----+------+---------------------+
| 5 | def | 2016-07-18 23:44:05 |
+----+------+---------------------+
1 row in set (0.00 sec) mysql> handler c close; //关闭
Query OK, 0 rows affected (0.00 sec)
mysql> handler test.test open;       //open
Query OK, 0 rows affected (0.00 sec) mysql> handler test read data first; //data索引,第一个记录
+----+------+---------------------+
| id | data | ts |
+----+------+---------------------+
| 1 | abc | 2016-07-18 23:44:05 |
+----+------+---------------------+
1 row in set (0.00 sec) mysql> handler test read data next; //下一个记录
+----+------+---------------------+
| id | data | ts |
+----+------+---------------------+
| 2 | abc | 2016-07-18 23:44:05 |
+----+------+---------------------+
1 row in set (0.00 sec) mysql> handler test read data prev; //前一个记录
+----+------+---------------------+
| id | data | ts |
+----+------+---------------------+
| 1 | abc | 2016-07-18 23:44:05 |
+----+------+---------------------+
1 row in set (0.01 sec) mysql> handler test read data last; //最后一条记录
+----+------+---------------------+
| id | data | ts |
+----+------+---------------------+
| 9 | yza | 2016-07-18 23:44:05 |
+----+------+---------------------+
1 row in set (0.00 sec)
mysql> handler test read `data`=("yza");
+----+------+---------------------+
| id | data | ts |
+----+------+---------------------+
| 9 | yza | 2016-07-18 23:44:05 |
+----+------+---------------------+
1 row in set (0.01 sec)
mysql> handler test read data=("abc") limit 5;
+----+------+---------------------+
| id | data | ts |
+----+------+---------------------+
| 1 | abc | 2016-07-18 23:44:05 |
| 2 | abc | 2016-07-18 23:44:05 |
+----+------+---------------------+
2 rows in set (0.00 sec)
mysql> handler test read first  limit 2;   //头两个记录  ID索引
+----+------+---------------------+
| id | data | ts |
+----+------+---------------------+
| 1 | abc | 2016-07-18 23:44:05 |
| 2 | abc | 2016-07-18 23:44:05 |
+----+------+---------------------+
2 rows in set (0.00 sec) mysql> handler test read first limit 3;
+----+------+---------------------+
| id | data | ts |
+----+------+---------------------+
| 1 | abc | 2016-07-18 23:44:05 |
| 2 | abc | 2016-07-18 23:44:05 |
| 3 | abd | 2016-07-18 23:44:05 |
+----+------+---------------------+
3 rows in set (0.15 sec) mysql> handler test read first ;
+----+------+---------------------+
| id | data | ts |
+----+------+---------------------+
| 1 | abc | 2016-07-18 23:44:05 |
+----+------+---------------------+
1 row in set (0.00 sec)
mysql> handler test read data last limit 5;   //data索引
+----+------+---------------------+
| id | data | ts |
+----+------+---------------------+
| 9 | yza | 2016-07-18 23:44:05 |
| 8 | vwx | 2016-07-18 23:44:05 |
| 7 | stu | 2016-07-18 23:44:05 |
| 6 | pqr | 2016-07-18 23:44:05 |
| 10 | def | 2016-07-18 23:44:05 |
+----+------+---------------------+
5 rows in set (0.00 sec)

Handler_read_first :

mysql> flush status;
Query OK, 0 rows affected (0.00 sec) mysql> show session status like "%handler%";
+----------------------------+-------+
| Variable_name | Value |
+----------------------------+-------+
| Handler_commit | 0 |
| Handler_delete | 0 |
| Handler_discover | 0 |
| Handler_external_lock | 0 |
| Handler_mrr_init | 0 |
| Handler_prepare | 0 |
| Handler_read_first | 0 |
| Handler_read_key | 0 |
| Handler_read_last | 0 |
| Handler_read_next | 0 |
| Handler_read_prev | 0 |
| Handler_read_rnd | 0 |
| Handler_read_rnd_next | 0 |
| Handler_rollback | 0 |
| Handler_savepoint | 0 |
| Handler_savepoint_rollback | 0 |
| Handler_update | 0 |
| Handler_write | 0 |
+----------------------------+-------+
18 rows in set (0.03 sec) mysql> handler test read data first;
+----+------+---------------------+
| id | data | ts |
+----+------+---------------------+
| 1 | abc | 2016-07-18 23:44:05 |
+----+------+---------------------+
1 row in set (0.00 sec) mysql> show session status like "%handler%";
+----------------------------+-------+
| Variable_name | Value |
+----------------------------+-------+
| Handler_commit | 1 |
| Handler_delete | 0 |
| Handler_discover | 0 |
| Handler_external_lock | 2 |
| Handler_mrr_init | 0 |
| Handler_prepare | 0 |
| Handler_read_first | 1 |
| Handler_read_key | 1 |
| Handler_read_last | 0 |
| Handler_read_next | 0 |
| Handler_read_prev | 0 |
| Handler_read_rnd | 0 |
| Handler_read_rnd_next | 0 |
| Handler_rollback | 0 |
| Handler_savepoint | 0 |
| Handler_savepoint_rollback | 0 |
| Handler_update | 0 |
| Handler_write | 0 |
+----------------------------+-------+
18 rows in set (0.02 sec) mysql> handler test read data first;
+----+------+---------------------+
| id | data | ts |
+----+------+---------------------+
| 1 | abc | 2016-07-18 23:44:05 |
+----+------+---------------------+
1 row in set (0.07 sec) mysql> show session status like "%handler%";
+----------------------------+-------+
| Variable_name | Value |
+----------------------------+-------+
| Handler_commit | 2 |
| Handler_delete | 0 |
| Handler_discover | 0 |
| Handler_external_lock | 4 |
| Handler_mrr_init | 0 |
| Handler_prepare | 0 |
| Handler_read_first | 2 |
| Handler_read_key | 2 |
| Handler_read_last | 0 |
| Handler_read_next | 0 |
| Handler_read_prev | 0 |
| Handler_read_rnd | 0 |
| Handler_read_rnd_next | 0 |
| Handler_rollback | 0 |
| Handler_savepoint | 0 |
| Handler_savepoint_rollback | 0 |
| Handler_update | 0 |
| Handler_write | 0 |
+----------------------------+-------+
18 rows in set (0.03 sec)

在Inside君的24C的测试服务器上,64线程主键查询跑到了近37W QPS,还是非常令人印象深刻的。对比SQL的SELECT查询,整体测试结果如下图所示:

HANDLER命令与实现

命令HANDLER的主要实现在源码sql_handler.h、sql_handler.cc,设个断点就能观察到具体的流程。MySQL上层及InnoDB存储引擎层主要实现函数入口为:

  • Sql_cmd_handler_open::execute
  • Sql_cmd_handler_read::execute
  • Sql_cmd_handler_close::execute
  • ha_innobase::init_table_handle_for_HANDLER
  • ha_partition::init_table_handle_for_HANDLER()(7版本支持HANDLER操作分区表)

既然性能不错,为什么在生产环境中并不见到命令HANDLER的使用呢?主要是因为HANDLER命令存在以下几个主要问题:

  1. 非一致性读取???
  2. 返回聚集索引中的所有列(即使是二级索引访问),而不能返回某个具体列
  3. 二级索引不使用LIMIT关键字,只能返回1行记录

知道命令HANDLER的同学,可能会认为HANDLER读取存在脏读问题。因为MySQL官方文档对于HANDLER读取的说明就是这么说的:

The handler interface does not have to provide a consistent look of the data (for example, dirty reads are permitted), so the storage engine can use optimizations that SELECT does not normally permit.

然而需要特别注意的是,MySQL文档中准确的说法是可以允许提供不一致的读取。但InnoDB存储引擎的HANDLER实现是支持一致性读取的,Inside君亲测的确不存在脏读问题。当然,源码说明一切,可以发现在函数init_table_handle_for_HANDLER会对READVIEW进行分配,而注释也说明了这点:

/* We let HANDLER always to do the reads as consistent reads, even
if the trx isolation level would have been specified as SERIALIZABLE */
m_prebuilt->select_lock_type = LOCK_NONE;
m_prebuilt->stored_select_lock_type = LOCK_NONE;

貌似用HANDLER命令来做主键的查询是不错的,减少了SQL解析器的开销,性能提升杠杠的。但为此,应用要付出巨大的改动,而SQL最大的优势就在于标准化。相信这也是目前NoSQL数据库遇到的最大的一个问题。比如MongoDB,Inside君每次写查询时都要打开官方的命令对照表……