* GreatSQL社区原创内容未经授权不得随意使用,转载请联系小编并注明来源。
一、const_table介绍
GreatSQL 的优化器主要用JOIN类来进行处理SQL语句的,JOIN类有以下四个table数量相关的成员变量。其中const_tables是optimize最开始就检查并且标识的,因为这样可以把记录最少的表放在执行计划的第一步,在后面的执行计划里面这些const tables是不参与循环遍历和计算的,因此可以减少很多开销。
计数名称 说明 哪个函数进行累加 tables 该查询语句的所有表的数量,包含物化表和临时表 JOIN::get_best_combination() primary_tables 该查询语句的主要表的数量,不包含物化表 JOIN::get_best_combination() const_tables 该查询语句中只有0行或者1行的表数量 JOIN::extract_const_tables和JOIN::extract_func_dependent_tables tmp_tables 该查询语句中临时表数量 JOIN::make_tmp_tables_info()
下面用一个简单的例子来说明 const_table 是什么。
greatsql> CREATE TABLE t1 (c1 int primary key, c2 varchar(32),date1 datetime);
greatsql> INSERT INTO t1 VALUES (1,'aaa','2021-03-25 16:44:00.123456'),(2,'bbb','2022-03-25 16:44:00.123456'),(3,'ccc','2023-03-25 16:44:00.123456');
# 这里看到以下的type类型是const,说明在优化器看来这是一张const table
greatsql> EXPLAIN SELECT * FROM t1 WHERE c1=1 AND c1<10;
+----+-------------+-------+------------+-------+---------------+---------+---------+-------+------+----------+-------+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+----+-------------+-------+------------+-------+---------------+---------+---------+-------+------+----------+-------+
| 1 | SIMPLE | t1 | NULL | const | PRIMARY | PRIMARY | 4 | const | 1 | 100.00 | NULL |
+----+-------------+-------+------------+-------+---------------+---------+---------+-------+------+----------+-------+
1 row in set, 1 warning (3.60 sec)
# 这句sql语句最后被处理成以下的语句了,可以看到条件语句被提到前面当做列,最后的条件变为where true了。
greatsql> SHOW WARNINGS;
+-------+------+-------------------------------------------------------------------------------------------------------------+
| Level | Code | Message |
+-------+------+-------------------------------------------------------------------------------------------------------------+
| Note | 1003 | /* select#1 */ select '1' AS `c1`,'aaa' AS `c2`,'2021-03-25 16:44:00' AS `date1` from `db1`.`t1` where true |
+-------+------+-------------------------------------------------------------------------------------------------------------+
1 row in set (0.00 sec)
二、const_table标识
const_table的寻找和标识在函数JOIN::extract_const_tables
和JOIN::extract_func_dependent_tables
进行处理,这两个函数的功能各有不同。用上面的例子来说明。
bool JOIN::make_join_plan() {
if (!(query_block->active_options() & OPTION_NO_CONST_TABLES)) {
// Detect tables that are const (0 or 1 row) and read their contents.
//根据表的统计信息table->file->stats.records直接寻找该值为0或者1的表,并且表的cached_table_flags必须为HA_STATS_RECORDS_IS_EXACT,只有引擎是MyISAM和memory才有这个标志
if (extract_const_tables()) return true;
// Detect tables that are functionally dependent on const values.
// 循环所有剩下的没有被标识的表,找出有索引或者有表依赖的表来标识
if (extract_func_dependent_tables()) return true;
}
}
bool JOIN::extract_func_dependent_tables() {
do {
for (JOIN_TAB **pos = best_ref + const_tables; *pos; pos++) {
一、遍历每张出现的表,寻找表里已经创建的索引。
二、判断表有join条件的情况,本次不涉及
三、判断表有表依赖的情况,本次不涉及
四、检查表是否可以通过key读取或表仅使用 const ref。
这里需要排除以下表:
1. 全文搜索,或
2. 嵌套外连接的一部分,或
3. 半连接的一部分,或
4. 具有大开销的外连接条件。
5. 被 const 表优化的处理程序阻止。
6. 将不会被使用,通常是因为它们是流式传输的而不是物化的(请参阅 Query_expression::can_materialize_directly_into_result())。
7. 位于完整连接的两侧
// 如果找到表有索引,这里用到的keyuse在之前JOIN::update_ref_and_keys()已经获取到。
if (eq_part.is_prefix(table->key_info[key].user_defined_key_parts)) {
// 该索引是唯一索引
if (table->key_info[key].flags & HA_NOSAME) {
if (const_ref == eq_part) {
// 标识该表是const table
mark_const_table(tab, start_keyuse);
// 改变table的状态为const
join_read_const_table(tab, positions + const_tables - 1);
}
}
}
}
} while ((const_table_map & found_ref) && ref_changed);
以下是实际使用的时候表扫描方式类型汇总,其中JT_CONST和JT_SYSTEM就是我们本章要说明的。JT_SYSTEM类型,这个需要跟JT_CONST区分开,JT_SYSTEM主要是扩展表而不是物理表,JT_SYSTEM表没有索引。JT_SYSTEM类型也会被计入const_tables,详情见下面。
join_type访问方法的类型 说明 JT_UNKNOWN 无效 JT_SYSTEM 表只有一行,比如select * from (select 1) JT_CONST 表最多只有一行满足,比如WHERE table.pk = 3 JT_EQ_REF '=符号用在唯一索引 JT_REF '=符号用在非唯一索引 JT_ALL 全表扫描 JT_RANGE 范围扫描 JT_INDEX_SCAN 索引扫描 JT_FT Fulltext索引扫描 JT_REF_OR_NULL 包含null值,比如"WHERE col = ... OR col IS NULL JT_INDEX_MERGE 一张表执行多次范围扫描最后合并结果
-- 下面的select 1表类型就是system而不是const。但是select 1这张表的数量也会计入const_tables
greatsql> EXPLAIN SELECT * FROM (SELECT 1);
+----+-------------+------------+------------+--------+---------------+------+---------+------+------+----------+----------------+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+----+-------------+------------+------------+--------+---------------+------+---------+------+------+----------+----------------+
| 1 | PRIMARY | <derived2> | NULL | system | NULL | NULL | NULL | NULL | 1 | 100.00 | NULL |
| 2 | DERIVED | NULL | NULL | NULL | NULL | NULL | NULL | NULL | NULL | NULL | No tables used |
+----+-------------+------------+------------+--------+---------------+------+---------+------+------+----------+----------------+
2 rows in set, 1 warning (12.33 sec)
三、实际例子说明
接下来看几个例子来说明上面的代码。
greatsql> CREATE TABLE t2 (c1 int, c3 varchar(32));
greatsql> CREATE index idx1 ON t2(c1);
greatsql> INSERT INTO t2 VALUES (1,'aaa'),(2,'bbb');
-- 非唯一索引不是const table
greatsql> EXPLAIN SELECT * FROM t2 WHERE c1=1 AND c1<10;
+----+-------------+-------+------------+------+---------------+------+---------+-------+------+----------+-------+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+----+-------------+-------+------------+------+---------------+------+---------+-------+------+----------+-------+
| 1 | SIMPLE | t2 | NULL | ref | idx1 | idx1 | 4 | const | 1 | 100.00 | NULL |
+----+-------------+-------+------------+------+---------------+------+---------+-------+------+----------+-------+
1 row in set, 1 warning (3.84 sec)
-- 改变c1为primary key,看一下结果。
-- between条件也是唯一条件,因此是const table
-- 这里有个约束条件:BETWEEN的上下界条件值必须一样,才会被判定为const,否则是range
greatsql> EXPLAIN SELECT * FROM t1 WHERE c1 BETWEEN 1 AND 1 AND c1<10;
+----+-------------+-------+------------+-------+---------------+---------+---------+-------+------+----------+-------+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+----+-------------+-------+------------+-------+---------------+---------+---------+-------+------+----------+-------+
| 1 | SIMPLE | t1 | NULL | const | PRIMARY | PRIMARY | 4 | const | 1 | 100.00 | NULL |
+----+-------------+-------+------------+-------+---------------+---------+---------+-------+------+----------+-------+
1 row in set, 1 warning (0.00 sec)
-- value in后面的值即使最小值和最大值相同也走的范围扫描,不被判定为const table
greatsql> EXPLAIN SELECT * FROM t1 WHERE c1 IN (1,1) AND c1<10;
+----+-------------+-------+------------+-------+---------------+---------+---------+------+------+----------+-------------+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+----+-------------+-------+------------+-------+---------------+---------+---------+------+------+----------+-------------+
| 1 | SIMPLE | t1 | NULL | range | PRIMARY | PRIMARY | 4 | NULL | 1 | 100.00 | Using where |
+----+-------------+-------+------------+-------+---------------+---------+---------+------+------+----------+-------------+
1 row in set, 1 warning (0.01 sec)
-- 下面的表只有一行,但是由于引擎不是MyISAM或者memory而不被判定为const table
greatsql> CREATE TABLE t3 AS SELECT 1 AS c1;
greatsql> EXPLAIN SELECT * FROM t3 WHERE c1=1;
+----+-------------+-------+------------+------+---------------+------+---------+------+------+----------+-------------+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+----+-------------+-------+------------+------+---------------+------+---------+------+------+----------+-------------+
| 1 | SIMPLE | t3 | NULL | ALL | NULL | NULL | NULL | NULL | 1 | 100.00 | Using where |
+----+-------------+-------+------------+------+---------------+------+---------+------+------+----------+-------------+
1 row in set, 1 warning (1.91 sec)
-- 下面的表只有一行,由于引擎是MEMORY而被判定为const table
greatsql> CREATE TABLE t5 engine=memory AS SELECT 1 AS c1;
greatsql> EXPLAIN SELECT * FROM t5 WHERE c1=1;
+----+-------------+-------+------------+--------+---------------+------+---------+------+------+----------+-------+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+----+-------------+-------+------------+--------+---------------+------+---------+------+------+----------+-------+
| 1 | SIMPLE | t5 | NULL | system | NULL | NULL | NULL | NULL | 1 | 100.00 | NULL |
+----+-------------+-------+------------+--------+---------------+------+---------+------+------+----------+-------+
1 row in set, 1 warning (2.70 sec)
-- 两张表进行join的时候,即使这里t1的顺序在后面,因为在优化器里面被判定为const table,所以顺序被提前到第一位。
greatsql> EXPLAIN SELECT t1.c1,t1.c2 FROM t2,t1 WHERE t1.c1=1 AND t1.c1<10;
+----+-------------+-------+------------+-------+---------------+---------+---------+-------+------+----------+-------------+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+----+-------------+-------+------------+-------+---------------+---------+---------+-------+------+----------+-------------+
| 1 | SIMPLE | t1 | NULL | const | PRIMARY | PRIMARY | 4 | const | 1 | 100.00 | NULL |
| 1 | SIMPLE | t2 | NULL | index | NULL | idx1 | 4 | NULL | 1 | 100.00 | Using index |
+----+-------------+-------+------------+-------+---------------+---------+---------+-------+------+----------+-------------+
2 rows in set, 1 warning (2.70 sec)
greatsql> EXPLAIN FORMAT=TREE SELECT t1.c1,t1.c2 FROM t2,t1 WHERE t1.c1=1 AND t1.c1<10;
+-----------------------------------------------------+
| EXPLAIN |
+-----------------------------------------------------+
| -> Index scan on t2 using idx1 (cost=0.35 rows=1) |
+-----------------------------------------------------+
1 row in set (40.27 sec)
-- 通过打印堆栈查看当前最佳的排列顺序,可以看到表t1确实被提到第一个了。
Thread 56 "connection" hit Breakpoint 16, JOIN::get_best_combination (this=0x7fff14c6d0f0)
at /sql/sql_optimizer.cc:3663
3663 for (Table_ref *sj_nest : query_block->sj_nests) {
(gdb) p best_ref[0]->table_ref->table_name
$99 = 0x7fff14c5d660 "t1"
(gdb) p best_ref[1]->table_ref->table_name
$100 = 0x7fff14c5d538 "t2"
四、总结
从上面优化器最早的步骤我们认识了const table的定义和判定方法,可以发现实际运用中const table是可以提高查询速度的。这里面通过唯一索引来定位查找数据是最快的就是因为被判定为const table,在后面的实践里面发现 =
(等值查询)、特殊情况下的 BETWEEN
的时候都可以被认为是const table,但是 IN
却不行,因此实际运用的时候尽量不要用IN
条件,而 多用等值条件 和 特殊条件下的BETWEEN
来查询,最好是只用等值条件。
一、const_table介绍
GreatSQL 的优化器主要用JOIN类来进行处理SQL语句的,JOIN类有以下四个table数量相关的成员变量。其中const_tables是optimize最开始就检查并且标识的,因为这样可以把记录最少的表放在执行计划的第一步,在后面的执行计划里面这些const tables是不参与循环遍历和计算的,因此可以减少很多开销。
计数名称 | 说明 | 哪个函数进行累加 |
---|---|---|
tables | 该查询语句的所有表的数量,包含物化表和临时表 | JOIN::get_best_combination() |
primary_tables | 该查询语句的主要表的数量,不包含物化表 | JOIN::get_best_combination() |
const_tables | 该查询语句中只有0行或者1行的表数量 | JOIN::extract_const_tables和JOIN::extract_func_dependent_tables |
tmp_tables | 该查询语句中临时表数量 | JOIN::make_tmp_tables_info() |
下面用一个简单的例子来说明 const_table 是什么。
greatsql> CREATE TABLE t1 (c1 int primary key, c2 varchar(32),date1 datetime);
greatsql> INSERT INTO t1 VALUES (1,'aaa','2021-03-25 16:44:00.123456'),(2,'bbb','2022-03-25 16:44:00.123456'),(3,'ccc','2023-03-25 16:44:00.123456');
# 这里看到以下的type类型是const,说明在优化器看来这是一张const table
greatsql> EXPLAIN SELECT * FROM t1 WHERE c1=1 AND c1<10;
+----+-------------+-------+------------+-------+---------------+---------+---------+-------+------+----------+-------+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+----+-------------+-------+------------+-------+---------------+---------+---------+-------+------+----------+-------+
| 1 | SIMPLE | t1 | NULL | const | PRIMARY | PRIMARY | 4 | const | 1 | 100.00 | NULL |
+----+-------------+-------+------------+-------+---------------+---------+---------+-------+------+----------+-------+
1 row in set, 1 warning (3.60 sec)
# 这句sql语句最后被处理成以下的语句了,可以看到条件语句被提到前面当做列,最后的条件变为where true了。
greatsql> SHOW WARNINGS;
+-------+------+-------------------------------------------------------------------------------------------------------------+
| Level | Code | Message |
+-------+------+-------------------------------------------------------------------------------------------------------------+
| Note | 1003 | /* select#1 */ select '1' AS `c1`,'aaa' AS `c2`,'2021-03-25 16:44:00' AS `date1` from `db1`.`t1` where true |
+-------+------+-------------------------------------------------------------------------------------------------------------+
1 row in set (0.00 sec)
二、const_table标识
const_table的寻找和标识在函数JOIN::extract_const_tables
和JOIN::extract_func_dependent_tables
进行处理,这两个函数的功能各有不同。用上面的例子来说明。
bool JOIN::make_join_plan() {
if (!(query_block->active_options() & OPTION_NO_CONST_TABLES)) {
// Detect tables that are const (0 or 1 row) and read their contents.
//根据表的统计信息table->file->stats.records直接寻找该值为0或者1的表,并且表的cached_table_flags必须为HA_STATS_RECORDS_IS_EXACT,只有引擎是MyISAM和memory才有这个标志
if (extract_const_tables()) return true;
// Detect tables that are functionally dependent on const values.
// 循环所有剩下的没有被标识的表,找出有索引或者有表依赖的表来标识
if (extract_func_dependent_tables()) return true;
}
}
bool JOIN::extract_func_dependent_tables() {
do {
for (JOIN_TAB **pos = best_ref + const_tables; *pos; pos++) {
一、遍历每张出现的表,寻找表里已经创建的索引。
二、判断表有join条件的情况,本次不涉及
三、判断表有表依赖的情况,本次不涉及
四、检查表是否可以通过key读取或表仅使用 const ref。
这里需要排除以下表:
1. 全文搜索,或
2. 嵌套外连接的一部分,或
3. 半连接的一部分,或
4. 具有大开销的外连接条件。
5. 被 const 表优化的处理程序阻止。
6. 将不会被使用,通常是因为它们是流式传输的而不是物化的(请参阅 Query_expression::can_materialize_directly_into_result())。
7. 位于完整连接的两侧
// 如果找到表有索引,这里用到的keyuse在之前JOIN::update_ref_and_keys()已经获取到。
if (eq_part.is_prefix(table->key_info[key].user_defined_key_parts)) {
// 该索引是唯一索引
if (table->key_info[key].flags & HA_NOSAME) {
if (const_ref == eq_part) {
// 标识该表是const table
mark_const_table(tab, start_keyuse);
// 改变table的状态为const
join_read_const_table(tab, positions + const_tables - 1);
}
}
}
}
} while ((const_table_map & found_ref) && ref_changed);
以下是实际使用的时候表扫描方式类型汇总,其中JT_CONST和JT_SYSTEM就是我们本章要说明的。JT_SYSTEM类型,这个需要跟JT_CONST区分开,JT_SYSTEM主要是扩展表而不是物理表,JT_SYSTEM表没有索引。JT_SYSTEM类型也会被计入const_tables,详情见下面。
join_type访问方法的类型 | 说明 |
---|---|
JT_UNKNOWN | 无效 |
JT_SYSTEM | 表只有一行,比如select * from (select 1) |
JT_CONST | 表最多只有一行满足,比如WHERE table.pk = 3 |
JT_EQ_REF | '=符号用在唯一索引 |
JT_REF | '=符号用在非唯一索引 |
JT_ALL | 全表扫描 |
JT_RANGE | 范围扫描 |
JT_INDEX_SCAN | 索引扫描 |
JT_FT | Fulltext索引扫描 |
JT_REF_OR_NULL | 包含null值,比如"WHERE col = ... OR col IS NULL |
JT_INDEX_MERGE | 一张表执行多次范围扫描最后合并结果 |
-- 下面的select 1表类型就是system而不是const。但是select 1这张表的数量也会计入const_tables
greatsql> EXPLAIN SELECT * FROM (SELECT 1);
+----+-------------+------------+------------+--------+---------------+------+---------+------+------+----------+----------------+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+----+-------------+------------+------------+--------+---------------+------+---------+------+------+----------+----------------+
| 1 | PRIMARY | <derived2> | NULL | system | NULL | NULL | NULL | NULL | 1 | 100.00 | NULL |
| 2 | DERIVED | NULL | NULL | NULL | NULL | NULL | NULL | NULL | NULL | NULL | No tables used |
+----+-------------+------------+------------+--------+---------------+------+---------+------+------+----------+----------------+
2 rows in set, 1 warning (12.33 sec)
三、实际例子说明
接下来看几个例子来说明上面的代码。
greatsql> CREATE TABLE t2 (c1 int, c3 varchar(32));
greatsql> CREATE index idx1 ON t2(c1);
greatsql> INSERT INTO t2 VALUES (1,'aaa'),(2,'bbb');
-- 非唯一索引不是const table
greatsql> EXPLAIN SELECT * FROM t2 WHERE c1=1 AND c1<10;
+----+-------------+-------+------------+------+---------------+------+---------+-------+------+----------+-------+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+----+-------------+-------+------------+------+---------------+------+---------+-------+------+----------+-------+
| 1 | SIMPLE | t2 | NULL | ref | idx1 | idx1 | 4 | const | 1 | 100.00 | NULL |
+----+-------------+-------+------------+------+---------------+------+---------+-------+------+----------+-------+
1 row in set, 1 warning (3.84 sec)
-- 改变c1为primary key,看一下结果。
-- between条件也是唯一条件,因此是const table
-- 这里有个约束条件:BETWEEN的上下界条件值必须一样,才会被判定为const,否则是range
greatsql> EXPLAIN SELECT * FROM t1 WHERE c1 BETWEEN 1 AND 1 AND c1<10;
+----+-------------+-------+------------+-------+---------------+---------+---------+-------+------+----------+-------+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+----+-------------+-------+------------+-------+---------------+---------+---------+-------+------+----------+-------+
| 1 | SIMPLE | t1 | NULL | const | PRIMARY | PRIMARY | 4 | const | 1 | 100.00 | NULL |
+----+-------------+-------+------------+-------+---------------+---------+---------+-------+------+----------+-------+
1 row in set, 1 warning (0.00 sec)
-- value in后面的值即使最小值和最大值相同也走的范围扫描,不被判定为const table
greatsql> EXPLAIN SELECT * FROM t1 WHERE c1 IN (1,1) AND c1<10;
+----+-------------+-------+------------+-------+---------------+---------+---------+------+------+----------+-------------+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+----+-------------+-------+------------+-------+---------------+---------+---------+------+------+----------+-------------+
| 1 | SIMPLE | t1 | NULL | range | PRIMARY | PRIMARY | 4 | NULL | 1 | 100.00 | Using where |
+----+-------------+-------+------------+-------+---------------+---------+---------+------+------+----------+-------------+
1 row in set, 1 warning (0.01 sec)
-- 下面的表只有一行,但是由于引擎不是MyISAM或者memory而不被判定为const table
greatsql> CREATE TABLE t3 AS SELECT 1 AS c1;
greatsql> EXPLAIN SELECT * FROM t3 WHERE c1=1;
+----+-------------+-------+------------+------+---------------+------+---------+------+------+----------+-------------+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+----+-------------+-------+------------+------+---------------+------+---------+------+------+----------+-------------+
| 1 | SIMPLE | t3 | NULL | ALL | NULL | NULL | NULL | NULL | 1 | 100.00 | Using where |
+----+-------------+-------+------------+------+---------------+------+---------+------+------+----------+-------------+
1 row in set, 1 warning (1.91 sec)
-- 下面的表只有一行,由于引擎是MEMORY而被判定为const table
greatsql> CREATE TABLE t5 engine=memory AS SELECT 1 AS c1;
greatsql> EXPLAIN SELECT * FROM t5 WHERE c1=1;
+----+-------------+-------+------------+--------+---------------+------+---------+------+------+----------+-------+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+----+-------------+-------+------------+--------+---------------+------+---------+------+------+----------+-------+
| 1 | SIMPLE | t5 | NULL | system | NULL | NULL | NULL | NULL | 1 | 100.00 | NULL |
+----+-------------+-------+------------+--------+---------------+------+---------+------+------+----------+-------+
1 row in set, 1 warning (2.70 sec)
-- 两张表进行join的时候,即使这里t1的顺序在后面,因为在优化器里面被判定为const table,所以顺序被提前到第一位。
greatsql> EXPLAIN SELECT t1.c1,t1.c2 FROM t2,t1 WHERE t1.c1=1 AND t1.c1<10;
+----+-------------+-------+------------+-------+---------------+---------+---------+-------+------+----------+-------------+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+----+-------------+-------+------------+-------+---------------+---------+---------+-------+------+----------+-------------+
| 1 | SIMPLE | t1 | NULL | const | PRIMARY | PRIMARY | 4 | const | 1 | 100.00 | NULL |
| 1 | SIMPLE | t2 | NULL | index | NULL | idx1 | 4 | NULL | 1 | 100.00 | Using index |
+----+-------------+-------+------------+-------+---------------+---------+---------+-------+------+----------+-------------+
2 rows in set, 1 warning (2.70 sec)
greatsql> EXPLAIN FORMAT=TREE SELECT t1.c1,t1.c2 FROM t2,t1 WHERE t1.c1=1 AND t1.c1<10;
+-----------------------------------------------------+
| EXPLAIN |
+-----------------------------------------------------+
| -> Index scan on t2 using idx1 (cost=0.35 rows=1) |
+-----------------------------------------------------+
1 row in set (40.27 sec)
-- 通过打印堆栈查看当前最佳的排列顺序,可以看到表t1确实被提到第一个了。
Thread 56 "connection" hit Breakpoint 16, JOIN::get_best_combination (this=0x7fff14c6d0f0)
at /sql/sql_optimizer.cc:3663
3663 for (Table_ref *sj_nest : query_block->sj_nests) {
(gdb) p best_ref[0]->table_ref->table_name
$99 = 0x7fff14c5d660 "t1"
(gdb) p best_ref[1]->table_ref->table_name
$100 = 0x7fff14c5d538 "t2"
四、总结
从上面优化器最早的步骤我们认识了const table的定义和判定方法,可以发现实际运用中const table是可以提高查询速度的。这里面通过唯一索引来定位查找数据是最快的就是因为被判定为const table,在后面的实践里面发现 =
(等值查询)、特殊情况下的 BETWEEN
的时候都可以被认为是const table,但是 IN
却不行,因此实际运用的时候尽量不要用IN
条件,而 多用等值条件 和 特殊条件下的BETWEEN
来查询,最好是只用等值条件。
《GreatSQL运维实战》视频课程
GreatSQL数据库是一款开源免费数据库,可在普通硬件上满足金融级应用场景,具有高可用、高性能、高兼容、高安全等特性,可作为MySQL或Percona Server for MySQL的理想可选替换。
⏩GitHub : https://github.com/GreatSQL/
🆙BiliBili : https://space.bilibili.com/1363850082
(对文章有疑问或见解可去社区官网提出哦~)
加入微信交流群 | 加入QQ交流群 |