将mysql非分区表转换为分区表
查看表的分布状况
mysql> select count(*) from  justin;
+----------+
| count(*) |
+----------+
|  5845246 |
+----------+
1 row in set (0.00 sec)
mysql> select month(create_time),count(*) from justin group by  month(create_time);
+-----------------------+----------+
| month(create_time) | count(*) |
+-----------------------+----------+
|                     1 |  1128520 |
|                    11 |  1574965 |
|                    12 |  3141750 |
+-----------------------+----------+
3 rows in set (6.93 sec)
考虑以create_time为分区键建立分区表
第一步  创建中间表，以主键id和分区列为联合主键
CREATE TABLE `temp_justin` (
  `id` bigint(1) NOT NULL AUTO_INCREMENT COMMENT '流水号,自增',
  `create_time` datetime DEFAULT NULL COMMENT '订单日志创建时间(建立索引)',
  PRIMARY KEY (`id`,`create_time`),
) ENGINE=MyISAM AUTO_INCREMENT=6000000 DEFAULT CHARSET=utf8;
表已经存在580多万记录并且不断在增长，因此中间表初始的id值设置成6000000
增加分区，以月为单位
alter table temp_justin partition by range(to_days(create_time))
(
partition p1012 values less than (to_days('2011-01-01')),
partition p1101 values less than (to_days('2011-02-01')),
partition p1102 values less than (to_days('2011-03-01')),
partition p1103 values less than (to_days('2011-04-01')),
partition p1104 values less than (to_days('2011-05-01')),
partition p1105 values less than (to_days('2011-06-01')),
partition p1106 values less than (to_days('2011-07-01')),
partition p1107 values less than (to_days('2011-08-01')),
partition p1108 values less than (to_days('2011-09-01')),
partition p1109 values less than (to_days('2011-10-01')),
partition p11010 values less than (to_days('2011-11-01')),
partition p11011 values less than (to_days('2011-12-01')),
partition p11012 values less than (to_days('2012-01-01'))
);
第二步 重命名表
Alter table justin rename to justin_bak_110113;
Alter table temp_justin rename to justin;
第三步 同步数据
Insert into justin select * from temp_justin;
表里已经存在将近600万条记录，如此批量导入数据会对数据库性能影响很大。
每一万条提交一次，sleep 2s ，53万数据总耗时2 min 39.67 sec。
mysql> create procedure cp_data()
    -> begin
    -> declare i int;
    -> set i=0;
    -> while i<60 do
    -> insert into justin
    -> select * from justin_bak_110113
    -> where id >= i*10000 and  id <(i+1)*10000;
    -> set i=i+1;
    -> select sleep(2);
    -> end while;
    -> end||
Query OK, 0 rows affected (0.04 sec)
mysql>
mysql> delete from justin;
    -> ||
Query OK, 0 rows affected (0.00 sec)
mysql> delimiter ;
mysql> call cp_data();
+----------+
| sleep(2) |
+----------+
|        0 |
+----------+
1 row in set (2 min 39.67 sec)
Query OK, 0 rows affected (2 min 39.67 sec)
mysql> select count(*) from justin;
+----------+
| count(*) |
+----------+
|   525031 |
+----------+
1 row in set (0.00 sec)
查看执行计划，使用了分区扫描
mysql> explain
    -> select count(*) from justin where create_time
    -> <='2011-01-13' and create_time>'2011-01-04';
+----+-------------+--------------------+-------+---------------+---------+---------+------+--