查看测试表结构：
mysql> show create table im_message \G
*************************** 1. row ***************************
Table: im_message
Create Table: CREATE TABLE `im_message` (
`id` int(11) NOT NULL AUTO_INCREMENT,
`from_id` varchar(40) COLLATE utf8_bin NOT NULL,
`from_type` tinyint(1) NOT NULL DEFAULT '0',
`to_id` varchar(40) COLLATE utf8_bin NOT NULL,
`to_type` tinyint(1) NOT NULL DEFAULT '0',
`content` varchar(2048) COLLATE utf8_bin DEFAULT '',
`create_date` bigint(20) NOT NULL,
`update_date` bigint(20) NOT NULL,
`message_id` varchar(40) COLLATE utf8_bin NOT NULL,
`is_sync` tinyint(1) DEFAULT '0' COMMENT '是否同步 0 未同步 1 已同步',
`is_read` tinyint(1) DEFAULT '0' COMMENT '是否已读 0 未读 1 已读',
`is_withdraw` tinyint(1) DEFAULT '0' COMMENT '是否撤回 0 未撤 1 已撤',
`is_lastest` tinyint(1) DEFAULT '0' COMMENT '是否是最新回话消息 0 不是 1是',
PRIMARY KEY (`id`),
UNIQUE KEY `uidx_message_id` (`message_id`),
KEY `idx_date` (`create_date`),
KEY `idx_from_id` (`from_id`),
KEY `idx_to_id` (`to_id`),
KEY `idx_is_sync` (`is_sync`),
KEY `idx_update_date` (`update_date`),
KEY `idx_fid_tid` (`from_id`,`to_id`)
) ENGINE=InnoDB AUTO_INCREMENT=13264365 DEFAULT CHARSET=utf8 COLLATE=utf8_bin ROW_FORMAT=DYNAMIC
1 row in set (0.00 sec)查看SQL执行效果：
select max(id) as "id" from im_message WHERE from_id='bd29879279c44672a2fdffcb8428b4d1' or to_id='bd29879279c44672a2fdffcb8428b4d1' group by from_id, to_id；
| 8547247 |
| 7152294 |
| 6897129 |
| 12874034 |
| 10011290 |
| 8027198 |
| 7852741 |
| 9960496 |
| 6059399 |
| 10860981 |
| 9963172 |
| 13253445 |
+----------+
27827 rows in set (0.91 sec)查看执行计划：
mysql> explain select max(id) as “id” from im_message WHERE from_id=’bd29879279c44672a2fdffcb8428b4d1′ or to_id=’bd29879279c44672a2fdffcb8428b4d1′ group by from_id, to_id;
+—-+————-+————+————+————-+———————————–+———————–+———+——+——–+———-+—————————————————————————————+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+—-+————-+————+————+————-+———————————–+———————–+———+——+——–+———-+—————————————————————————————+
| 1 | SIMPLE | im_message | NULL | index_merge | idx_from_id,idx_to_id,idx_fid_tid | idx_fid_tid,idx_to_id | 122,122 | NULL | 168680 | 100.00 | Using sort_union(idx_fid_tid,idx_to_id); Using where; Using temporary; Using filesort |
+—-+————-+————+————+————-+———————————–+———————–+———+——+——–+———-+—————————————————————————————+
1 row in set, 1 warning (0.00 sec)
创建覆盖索引：
mysql> alter table im_message add index idx_from_id_id(from_id,id);
Query OK, 0 rows affected (1 min 1.94 sec)
Records: 0 Duplicates: 0 Warnings: 0
mysql> alter table im_message add index idx_to_id_id(to_id,id);
Query OK, 0 rows affected (1 min 9.79 sec)
Records: 0 Duplicates: 0 Warnings: 0重新查看SQL执行效果：
select max(id) as "id" from im_message WHERE from_id='bd29879279c44672a2fdffcb8428b4d1' or to_id='bd29879279c44672a2fdffcb8428b4d1' group by from_id, to_id;
| 8547247 |
| 7152294 |
| 6897129 |
| 12874034 |
| 10011290 |
| 8027198 |
| 7852741 |
| 9960496 |
| 6059399 |
| 10860981 |
| 9963172 |
| 13253445 |
+----------+
27827 rows in set (0.63 sec)查看执行计划：
mysql> explain select max(id) as "id" from im_message WHERE from_id='bd29879279c44672a2fdffcb8428b4d1' or to_id='bd29879279c44672a2fdffcb8428b4d1' group by from_id, to_id;
+----+-------------+------------+------------+-------------+---------------------------------------------------------------+-----------------------------+---------+------+--------+----------+----------------------------------------------------------------------------------------+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+----+-------------+------------+------------+-------------+---------------------------------------------------------------+-----------------------------+---------+------+--------+----------+----------------------------------------------------------------------------------------+
| 1 | SIMPLE | im_message | NULL | index_merge | idx_from_id,idx_to_id,idx_fid_tid,idx_from_id_id,idx_to_id_id | idx_from_id_id,idx_to_id_id | 122,122 | NULL | 162106 | 100.00 | Using union(idx_from_id_id,idx_to_id_id); Using where; Using temporary; Using filesort |
+----+-------------+------------+------------+-------------+---------------------------------------------------------------+-----------------------------+---------+------+--------+----------+----------------------------------------------------------------------------------------+
1 row in set, 1 warning (0.00 sec)发现优化器选择了新建的两个覆盖索引。
创建覆盖索引之后，利用索引的有序性， select max(id)可以快速的取到最大id。