MySQL分区

mysql中数据库learn目录结构:

看一下表sales的定义：

show create  table sales \G

*************************** 1. row ***************************
Table: sales
Create Table: CREATE TABLE `sales` (
  `id` int(11) NOT NULL AUTO_INCREMENT,
  `amount` double NOT NULL,
  `order_day` datetime NOT NULL,
  PRIMARY KEY (`id`,`order_day`)
) ENGINE=InnoDB DEFAULT CHARSET=latin1
/*!50100 PARTITION BY RANGE (YEAR(order_day))
(PARTITION p_2010 VALUES LESS THAN (2010) ENGINE = InnoDB,
 PARTITION p_2011 VALUES LESS THAN (2011) ENGINE = InnoDB,
 PARTITION p_2012 VALUES LESS THAN (2012) ENGINE = InnoDB,
 PARTITION p_catchall VALUES LESS THAN MAXVALUE ENGINE = InnoDB) */
1 row in set (0.00 sec)

表p_key的定义

show create  table p_key \G

*************************** 1. row ***************************
Table: p_key
Create Table: CREATE TABLE `p_key` (
  `id` int(10) NOT NULL AUTO_INCREMENT,
  `keyname` char(20) DEFAULT NULL,
  `keyval` varchar(1000) DEFAULT NULL,
  PRIMARY KEY (`id`)
) ENGINE=MyISAM AUTO_INCREMENT=12 DEFAULT CHARSET=utf8
/*!50100 PARTITION BY KEY (id)
PARTITIONS 4 */
1 row in set (0.01 sec)

对于MyISAM引擎，一张表对于存储了3个文件，fm存储表结构，myi存放索引，myd存放数据。但p_key对应的还有一个文件p_key.par。

重新创建一个Range分区的表fuhui_log，体验分区查询：

DROP TABLE IF EXISTS fuhui_log;
CREATE TABLE fuhui_log (
    object_id int(11),
    title varchar(20) NOT NULL  ,
    content varchar(20) ,
    time int(11),
    primary key (object_id)
)
PARTITION BY range (object_id)
(
    PARTITION p1 VALUES less than (5000),
    PARTITION p2 VALUES less than (10000),
    PARTITION p3 VALUES less than MAXVALUE
);

自定义存储过程，向数据库中插入20000条数据：

delimiter //
create procedure fun_fuhui_log() 
begin
    declare i int;
    set i = 1;
    while i < 20000 do
        insert into fuhui_log(object_id,title,content,time) values (i,concat(‘title_‘,i),‘test content‘,i);
        set i = i+1;
    end while;
end
//

调用存储过程，进行数据插入：

delimiter ;
call fun_fuhui_log();

获取插入数据结果：

 select count(*) from fuhui_log;

查询结果为19999，耗时：1 row in set (0.01 sec)；

select * from fuhui_log where object_id = 13588;

耗时0.00 sec

根据如上的步骤，创建一个基本表，并修改存储过程，插入同样的数据：

DROP TABLE IF EXISTS fuhui_log2;
CREATE TABLE fuhui_log2 (
    object_id int(11),
    title varchar(20) NOT NULL  ,
    content varchar(20) ,
    time int(11),
    primary key (object_id)
);

数据结构设计的太简单，数据量太小，看不出效果来，重先修改存储过程，插入80000条数据：

while i < 80000 do
        replace into fuhui_log2(object_id,title,content,time) values (i,concat(‘title_‘,i),‘test content‘,i);
        set i = i+1;
end while;

select count(*) from fuhui_log2;

执行结果：1 row in set (0.02 sec)

select count(*) from fuhui_log;

执行结果：1 row in set (0.03 sec)【没有按照逻辑出牌】

这个例子很失败，修改表结构，去掉primary key

 alter table fuhui_log drop primary key;
 alter table fuhui_log2 drop primary key;

例子仍然比较失败，执行的效率很难发现

select * from fuhui_log where object_id = 56770 \G

耗时：0.05sec

select * from fuhui_log2 where object_id = 56770 \G

耗时0.06sec

对于count统计，fuhui_log比fuhui_log2耗时都多，count的并行计算，都被我给玷污了

修改分区结构，重新计算：

 alter table fuhui_log reorganize partition p3 into (
 partition p3_1 values less than (30000),
 partition p3_2 values less than (50000),
 partition p3_3 values less than MAXVALUE);

查看重新分区后的结果：

select table_schema,table_name,partition_name,PARTITION_METHOD from infor
mation_schema.partitions where table_name=‘fuhui_log‘;

然后重新计算：

select count(*) from fuhui_log ;

执行效果0.04sec，跟fuhui_log2的统计时间相等了。但是

select * from fuhui_log where object_id = 56770 \G

执行时间变成了0.02sec

居然已经写这么久了，今天就此罢笔吧