首页 > 数据库 > MySQL > 正文

MySQL count 之索引挑选

2024-07-24 12:34:52
字体:
来源:转载
供稿:网友
  覆盖索引对于一些统计问题,如下:
 
  MySQL > show create table test1 /G
  *************************** 1. row ***************************
         Table: test1
  Create Table: CREATE TABLE `test1` (
    `id` bigint(16) NOT NULL AUTO_INCREMENT,
    `order_seq` bigint(16) NOT NULL,
    PRIMARY KEY (`id`),
    KEY `idx_id` (`id`),
    KEY `idx_id_ordseq` (`id`,`order_seq`)
  ) ENGINE=InnoDB AUTO_INCREMENT=15002212 DEFAULT CHARSET=utf8mb4
  1 row in set (0.00 sec)
  MySQL > explain select count(*) from test1 where id>10000 and id<20000;
  +----+-------------+-------+------------+-------+------------------------------+--------+---------+------+------+----------+--------------------------+
  | id | select_type | table | partitions | type  | possible_keys                | key    | key_len | ref  | rows | filtered | Extra                    |
  +----+-------------+-------+------------+-------+------------------------------+--------+---------+------+------+----------+--------------------------+
  |  1 | SIMPLE      | test1 | NULL       | range | PRIMARY,idx_id,idx_id_ordseq | idx_id | 8       | NULL | 9999 |   100.00 | Using where; Using index |
  +----+-------------+-------+------------+-------+------------------------------+--------+---------+------+------+----------+--------------------------+
  1 row in set, 1 warning (0.00 sec)
  innodb存储引擎选择了id字段的辅助索引,而不是聚集索引来统计,更不是联合索引。原因是辅助索引远小于聚集索引,选择辅助索引可以减少IO资源消耗。
 
  而另外一个统计场景:
 
  select count(*) from test1 where order_seq > 1502131212577 and order_seq< 202007080947244761;
 
  test1表建有id和 order_seq 字段的联合索引。
 
  MySQL > show create table test1 /G
  *************************** 1. row ***************************
         Table: test1
  Create Table: CREATE TABLE `test1` (
    `id` bigint(16) NOT NULL AUTO_INCREMENT,
    `order_seq` bigint(16) NOT NULL,
    PRIMARY KEY (`id`),
    KEY `idx_id` (`id`),
    KEY `idx_id_ordseq` (`id`,`order_seq`)
  ) ENGINE=InnoDB AUTO_INCREMENT=15002212 DEFAULT CHARSET=utf8mb4
  1 row in set (0.00 sec)
  MySQL > explain select count(*) from test1 where order_seq > 1502131212577 and order_seq< 202007080947244761;
  +----+-------------+-------+------------+-------+---------------+---------------+---------+------+----------+----------+--------------------------+
  | id | select_type | table | partitions | type  | possible_keys | key           | key_len | ref  | rows     | filtered | Extra                    |
  +----+-------------+-------+------------+-------+---------------+---------------+---------+------+----------+----------+--------------------------+
  |  1 | SIMPLE      | test1 | NULL       | index | NULL          | idx_id_ordseq | 16      | NULL | 15068082 |    11.11 | Using where; Using index |
  +----+-------------+-------+------------+-------+---------------+---------------+---------+------+----------+----------+--------------------------+
  1 row in set, 1 warning (0.00 sec)
  这里使用条件 order_seq 查询,一般情况下使用不了联合索引的,但是这个案例中的查询,利用到覆盖索引的信息。possible_keys依然为null,但是key是idx_id_ordseq,extra里出现Using index,表示为覆盖索引。

(编辑:武林网)

发表评论 共有条评论
用户名: 密码:
验证码: 匿名发表