SQL SERVER 2005 分区表实际操作( 2008-03-03 16:25:30| 分类： 数据库技术 )

     因为在中山IOD项目开发中,遇到数据量非常大的情况,一天一个表一个G,考虑数据查询效率问题,需要进行表分区的操作,具体的原理就类似于把一个表的资 料放在不同的分区里面,当查询时,如果都在同一个单独的分区内的话,就不用进行全表扫描,因此在这些情况下可以提高查询的效率,但如果所有分区都要查询所 有资料的话,分区并不会提高效率.

  下面是具体的SQL语句:

     1、最基本,最重要的一步就是创建分区函数.创建分区函数首先要确定分区键--既按照哪字段来进行分区.在这个例子里,我用提交时间来作为分区键,由于数据量的问题,最终决定每个月的数据放一个单独的分区.

         CREATE PARTITION FUNCTION IODPFN(VARCHAR(20))
                AS  RANGE LEFT FOR VALUES (
                           '2007/12/01 00:00:00',
                           '2008/01/01 00:00:00',
                           '2008/02/01 00:00:00',
                           '2008/03/01 00:00:00',
                            '2008/04/01 00:00:00',
                            '2008/05/01 00:00:00'
                            );

     2、上一步是完成一个概念上的分区,接下来要完成一个物理的构建,使得属于不同分区的数据存储到不同的物理文件上去.
      a.创建文件组

                ALTER DATABASE IOD ADD FILEGROUP [iod200712];
                ALTER DATABASE IOD ADD FILEGROUP [iod200801];
                ALTER DATABASE IOD ADD FILEGROUP [iod200802];
                ALTER DATABASE IOD ADD FILEGROUP [iod200803];
                 ALTER DATABASE IOD ADD FILEGROUP [iod200804];
                 ALTER DATABASE IOD ADD FILEGROUP [iod200805];

   b.创建物理文件,在这里,为了方便起见,我把每个物理文件放到了一个单独的文件组里面.

                ALTER DATABASE IOD ADD FILE
(NAME = 'iod200712',FILENAME = 'D:\Program Files\Microsoft SQL Server\MSSQL.2\MSSQL\Data\iod200712.ndf',SIZE = 1GB,MAXSIZE = UNLIMITED,FILEGROWTH = 100MB) TO FILEGROUP [iod200712];

               ALTER DATABASE IOD ADD FILE
(NAME = 'iod200801',FILENAME = 'D:\Program Files\Microsoft SQL Server\MSSQL.2\MSSQL\Data\iod200801.mdf',SIZE = 1GB,MAXSIZE = UNLIMITED,FILEGROWTH = 100MB) TO FILEGROUP [iod200801];

              ALTER DATABASE IOD ADD FILE
(NAME = 'iod200802',FILENAME = 'D:\Program Files\Microsoft SQL Server\MSSQL.2\MSSQL\Data\iod200802.ndf',SIZE = 1GB,MAXSIZE = UNLIMITED,FILEGROWTH = 100MB) TO FILEGROUP [iod200802];

             ALTER DATABASE IOD ADD FILE
(NAME = 'iod200803',FILENAME = 'D:\Program Files\Microsoft SQL Server\MSSQL.2\MSSQL\Data\iod200803.mdf',SIZE = 1GB,MAXSIZE = UNLIMITED,FILEGROWTH = 100MB) TO FILEGROUP [iod200803];

            ALTER DATABASE IOD ADD FILE
(NAME = 'iod200804',FILENAME = 'D:\Program Files\Microsoft SQL Server\MSSQL.2\MSSQL\Data\iod200804.mdf',SIZE = 1GB,MAXSIZE = UNLIMITED,FILEGROWTH = 100MB) TO FILEGROUP [iod200804];

           ALTER DATABASE IOD ADD FILE
(NAME = 'iod200805',FILENAME = 'D:\Program Files\Microsoft SQL Server\MSSQL.2\MSSQL\Data\iod200805.mdf',SIZE = 1GB,MAXSIZE = UNLIMITED,FILEGROWTH = 100MB) TO FILEGROUP [iod200805];

 

  3、创建完分区函数,接下来就要建立分区架构,用来将概念上的分区和文件组(物理文件)关联起来.

         CREATE PARTITION SCHEME IODPScheme
          AS  PARTITION IODPFN
        TO ([iod200712],[iod200801],[iod200802],[iod200803],[iod200804],[iod200805],[PRIMARY]);

4、最后是建表

        drop table iod_data;

       create table iod_data
(
id                   varchar(20) primary key,
msg_type             varchar(10),            
send_service_type    varchar(10),
receive_service_type varchar(10),         
src                  varchar(20),   
src_top              varchar(10),    
src_npi              varchar(10),
dest                 varchar(20),
dest_top             varchar(10),
dest_npi             varchar(10),
last_time            varchar(20),      
remote_user_type     varchar(10),
pid                  varchar(10),
dcs                  varchar(10),
msg_length           varchar(10),
msg_status           varchar(10),
send_ip              varchar(15),
resive_ip            varchar(15),
smid                 varchar(10),   
sm_status            varchar(10),    
sm_errcode           varchar(10),
fee                  varchar(10),
submit_time          varchar(20),
business_subtype     varchar(10),
msgid_timestamp      varchar(10),
msgid_gateway        varchar(10),
msgid_seq            varchar(10),
same_msg_sum         varchar(10),  
same_msg_seq         varchar(10),  
msg_level            varchar(10),
fee_user_type        varchar(10),
fee_terminal_Id      varchar(20),
msg_src              varchar(10),
fee_type             varchar(10),
fee_code             varchar(10),
service_code         varchar(20),
isnot_write_fee      varchar(10),  
confirm_type         varchar(10),
pps                  varchar(10),
local_gateway_code   varchar(10),
process_time         varchar(20),
src_gateway_code     varchar(10),
front_gateway_code   varchar(10),
content              varchar(255)
)
on IODPScheme(submit_time);

其它,可以建索引以提高查询速度,但插入数据的速度会变慢

  CREATE INDEX iod_data_ix1 ON iod_data(src,dest);
  CREATE INDEX iod_data_ix2 ON iod_data(fee_user_type);
  CREATE INDEX iod_data_ix3 ON iod_data(fee_type);