Oracle-分区表案例

Oracle数据库开发了解分区表
分区就是将一个非常大的表或者索引物理地分解为多个较小的 可独立管理的部分.
分区表或索引在逻辑上是一个表或一个索引,但物理上是由多个物理分区组成的.
分区功能通过改善可管理性 性能 可用性,为各种应用系统带来了极大的好处.
分区功能的好处:
 1.增强数据可用性:如果表的一个分区因故障或者维护而不能使用时,表的其余分区仍是可用的;
 2.维护方便:独立管理多个分区,比维护单个大表要轻松;
 3.均衡I/O:可以把不同分区映射到磁盘以平衡I/O,显著改善性能;
 4.改善查询性能:对已分区对象的某些查询可以运行更快,因为搜索仅限于关心的分区;
 
分区表有哪些??
 Oracle 11g 提供6种表分区方法:范围分区(range) 散列分区(hash) 列表分区(list) 
 符合分区 间隔分区 引用分区.
 
<范围分区 partition by range>
 按表中某个列值的范围进行分区,根据该列的值决定将数据存储在哪个分区上.
 
创建范围分区需注意以下几点:
 1.指明分区方法,分区列,和分区描述
 2.每一个分区都有values less than子句
 3.在最高分区中定义maxvalue,这个maxvalue值高区其他分区中的任何键值.
 

例:创建范围分区

create table range_orders
 (order_id varchar2(10) constraint OR_PK  primary key
 ,order_date date default sysdate
 ,qty integer
 ,payterms varchar2(10)
 ,book_id number(6)
 )
 partition by range (order_date)
 (partition p1 values less than (to_date('20140331','yyyymmdd')) tablespace user01,
  partition p2 values less than (to_date('20140430','yyyymmdd')) tablespace user02,
  partition p3 values less than (to_date('20140531','yyyymmdd')) tablespace user03
 )
 ;
 
SQL> insert into range_orders values ('10001',to_date('20140321','yyyymmdd'),1,'payterm_1',110345);
 1 row inserted
 SQL> insert into range_orders values ('10002',to_date('20140421','yyyymmdd'),1,'payterm_2',110745); 
 1 row inserted
 SQL> insert into range_orders values ('10003',to_date('20140521','yyyymmdd'),1,'payterm_3',110945);
 1 row inserted
 
SQL> commit;
 Commit complete
 
SQL> select rowid,r.* from range_orders r;
 ROWID              ORDER_ID  ORDER_DATE   QTY PAYTERMS  BOOK_ID
 ------------------ ---------- ----------- ---- ---------- -------
 AAADwpAAGAAAACFAAA 10001      2014/3/21     1 payterm_1  110345
 AAADwqAAHAAAACFAAA 10002      2014/4/21     1 payterm_2  110745
 AAADwrAAIAAAACFAAA 10003      2014/5/21     1 payterm_3  110945

可以看到AAG,AAH,AAI 分别代表了三条数据的文件号是6,7,8
备注:这个地方可以看下http://blog.itpub.net/28929558/viewspace-1150766/ 了解rowid
验证下

 SQL> select x.FILE#,x.NAME from v$datafile x;

      FILE# NAME
 ---------- ----------------------------------------
          1 D:\ORACLE\ORADATA\CRISS_DB\SYSTEM01.DBF
          2 D:\ORACLE\ORADATA\CRISS_DB\SYSAUX01.DBF
          3 D:\ORACLE\ORADATA\CRISS_DB\UNDOTBS01.DBF
          4 D:\ORACLE\ORADATA\CRISS_DB\USERS01.DBF
          5 D:\ORACLE\ORADATA\CRISS_DB\TEST01.DBF
          6 D:\ORACLE\ORADATA\CRISS_DB\USER01.DBF
          7 D:\ORACLE\ORADATA\CRISS_DB\USER02.DBF
          8 D:\ORACLE\ORADATA\CRISS_DB\USER03.DBF
  
 8 rows selected

<散列分区 partition by hash>
 散列分区指一个或多个列上应用一个散列函数,数根据该散列值存放在不同的分区中.
 通过散列分区,可以将数据比较均匀地分布到各个分区中.
 
例:创建散列分区表

SQL> create table hash_orders
	(order_id varchar2(10) constraint HOR_PK  primary key
	,order_date date default sysdate
	,qty integer
	,payterms varchar2(10)
	,book_id number(6)
	)
	partition by hash(order_id)
	( partition hash_p1 tablespace user01
	,partition hash_p2 tablespace user02
	);
  
 Table created
 
SQL> insert into hash_orders select * from range_orders;
  
 3 rows inserted

SQL> select rowid,h.* from hash_orders h;
  
 ROWID              ORDER_ID  ORDER_DATE    QTY PAYTERMS  BOOK_ID
 ------------------ ---------- ----------------- ---------- -------
 AAADyIAAGAAAACNAAA 10002      2014/4/21      1 payterm_2  110745
 AAADyIAAGAAAACNAAB 10003      2014/5/21      1 payterm_3  110945
 AAADyJAAHAAAACNAAA 10001      2014/3/21      1 payterm_1  110345

order_id为 10002,10003的被存在AAG 6号文件上,第141个块上的 第1行和第2行
           10001 存放在AAH 7号文件上
 
<列表分区 partition by list>
 当表中某列的值只有几个的时候,可以采用列表分区,即指定在几个(根据列值个数)表空间中.

SQL> create table list_orders
	(order_id varchar2(10) constraint LOR_PK  primary key
	,order_date date default sysdate
	,qty integer
	,payterms varchar2(10)
	,book_id number(6)
	)
	partition by list(payterms)
	( partition list_p1 values('payterm_1') tablespace user01
	,partition list_p2 values('payterm_2')tablespace user02
	,partition list_p3 values('payterm_3') tablespace user03
	);
  
 Table created
 
SQL> select rowid,l.* from list_orders l;
  
 ROWID              ORDER_ID  ORDER_DATE     QTY PAYTERMS  BOOK_ID
 ------------------ ---------- ------------------ ---------- -------
 AAADyRAAGAAAACdAAA 10001      2014/3/21       1 payterm_1  110345
 AAADySAAHAAAACdAAA 10002      2014/4/21       1 payterm_2  110745
 AAADyTAAIAAAACVAAA 10003      2014/5/21       1 payterm_3  110945

<复合分区>
 基于范围分区和列表分区的组合 或 范围分区和散列分区的组合

SQL> create table comp_orders
	(order_id varchar2(10) constraint COR_PK  primary key
	,order_date date default sysdate
	,qty integer
	,payterms varchar2(10)
	,book_id number(6)
	)
	partition by range (order_date)
	subpartition by list(payterms)
	(partition p1 values less than (to_date('20140331','yyyymmdd'))
			( subpartition p1_sub1 values('payterm_1') tablespace user01
				,subpartition p1_sub2 values('payterm_2') tablespace user02
				,subpartition p1_sub3 values('payterm_3') tablespace user03
				)
	,partition p2 values less than (to_date('20140430','yyyymmdd'))
			( subpartition p2_sub1 values('payterm_1') tablespace user04
				,subpartition p2_sub2 values('payterm_2') tablespace user05
				,subpartition p2_sub3 values('payterm_3') tablespace user06
				)
	,partition p3 values less than (maxvalue) tablespace user07
	)
	;
  
 Table created
 

SQL> select rowid,c.* from comp_orders c;
  
 ROWID              ORDER_ID  ORDER_DATE    QTY PAYTERMS  BOOK_ID
 ------------------ ---------- ----------- ----- ---------- -------
 AAADyZAAGAAAAClAAA 10004      2014/3/21      1 payterm_1  110345
 AAADyZAAGAAAAClAAB 10001      2014/3/21      1 payterm_1  110345
 AAADydAAKAAAACFAAA 10005      2014/4/21      1 payterm_2  110745
 AAADydAAKAAAACFAAB 10002      2014/4/21      1 payterm_2  110745
 AAADyfAAMAAAACFAAA 10006      2014/5/21      1 payterm_3  110945
 AAADyfAAMAAAACFAAB 10003      2014/5/21      1 payterm_3  110945
  
 6 rows selected

PAYTERMS字段值相同,且ORDER_DATE时间区间相同的数据被放到同一个表空间下面.


下面对比下PAYTERMS字段值相同,ORDER_DATE时间区间不同的结果

SQL> insert into comp_orders values('10007',to_date('20140421','yyyymmdd'),1,'payterm_1','113888');
 1 row inserted
  
 SQL> select rowid,c.* from comp_orders c;
  
 ROWID              ORDER_ID  ORDER_DATE   QTY PAYTERMS  BOOK_ID
 ------------------ ---------- ---------------- ---------- -------
 AAADyZAAGAAAAClAAA 10004      2014/3/21     1 payterm_1  110345
 AAADyZAAGAAAAClAAB 10001      2014/3/21     1 payterm_1  110345                                   
 AAADycAAJAAAACFAAA 10007      2014/4/21     1 payterm_1  113888                            
 AAADydAAKAAAACFAAA 10005      2014/4/21     1 payterm_2  110745
 AAADydAAKAAAACFAAB 10002      2014/4/21     1 payterm_2  110745
 AAADyfAAMAAAACFAAA 10006      2014/5/21     1 payterm_3  110945
 AAADyfAAMAAAACFAAB 10003      2014/5/21     1 payterm_3  110945
  
 7 rows selected

ORDER_ID为10007的数据存放在了另一个表空间内(AAJ AAADyc[AAJ]AAAACFAAA)

<间隔分区>
 间隔分区是oracle 11g release 1 以后版本中新增的特性,它是对范围分区的扩展,
 可以自动进行等距离范围的分区.
 
实验对比一下,间隔分区与范围分区有什么区别即可.
 
首先创建一个间隔分区,间隔分区以一个范围分区为'起点',并定义一个间隔,
当有数据插入,依据该间隔为附加的数据创建新的分区.

SQL> create table intvl_orders
	(order_id varchar2(10) constraint INO_PK  primary key
	,order_date date default sysdate
	,qty integer
	,payterms varchar2(10)
	,book_id number(6)
	)
	partition by range (order_date)
	interval(numtoyminterval(1,'MONTH'))
	store in (user01,user02,user03)
	(
	partition p1 values less than (to_date('20140101','yyyymmdd'))
	)
	;
  
 Table created

注意:partition p1 values less than (to_date('20140101','yyyymmdd')) 起始是月初,
不然会报错 ORA-14767: 无法使用现有上限指定此间隔.至于为什么,大家自行去查找下答案吧.
  
 重新创建一个范围分区表并插入数据！

SQL> drop table range_orders purge;
  
 Table dropped
  
 SQL> 
 SQL> create table range_orders
	(order_id varchar2(10) constraint OR_PK  primary key
	,order_date date default sysdate
	,qty integer
	,payterms varchar2(10)
	,book_id number(6)
	)
	partition by range (order_date)
	(partition p1 values less than (to_date('20140331','yyyymmdd')) tablespace user01,
	partition p2 values less than (to_date('20140430','yyyymmdd')) tablespace user02,
	partition p3 values less than (to_date('20140531','yyyymmdd')) tablespace user03
	)
	;
  
 Table created
  
 SQL> insert into range_orders values ('10003',to_date('20140621','yyyymmdd'),1,'payterm_3',110945);
  
 insert into range_orders values ('10003',to_date('20140621','yyyymmdd'),1,'payterm_3',110945)
  
 ORA-14400: 插入的分区关键字未映射到任何分区
  
 SQL> insert into range_orders values ('10003',to_date('20140521','yyyymmdd'),1,'payterm_3',110945);
  
 1 row inserted

可以看到,新创建的范围分区表最大范围是20140531,当出入的数据超出分区范围就会报错,
需要手工去做分区扩展.下面再给间隔分区插入数据

SQL> insert into intvl_orders values ('10003',to_date('20140621','yyyymmdd'),1,'payterm_3',110945);
  
 1 row inserted

我们插入同样超出分区范围的数据,数据插入成功了!
 
我们看一下前后两次的建表语句有什么变化
刚刚建好间隔分区表时候oracle内的建表语句:

-- Create table
 create table INTVL_ORDERS
 (
  ORDER_ID  VARCHAR2(10) not null,
  ORDER_DATE DATE default sysdate,
  QTY        INTEGER,
  PAYTERMS  VARCHAR2(10),
  BOOK_ID    NUMBER(6)
 )
 partition by range (ORDER_DATE)
 (
  partition P1 values less than (TO_DATE(' 2014-01-01 00:00:00', 'SYYYY-MM-DD HH24:MI:SS', 'NLS_CALENDAR=GREGORIAN'))
    tablespace USERS
    pctfree 10
    initrans 1
    maxtrans 255
    storage
    (
      initial 64K
      next 1M
      minextents 1
      maxextents unlimited
    )
 );
 -- Create/Recreate primary, unique and foreign key constraints 
 alter table INTVL_ORDERS
  add constraint INO_PK primary key (ORDER_ID)
  using index 
  tablespace USERS
  pctfree 10
  initrans 2
  maxtrans 255
  storage
  (
    initial 64K
    next 1M
    minextents 1
    maxextents unlimited
  );

插入数据后间隔分区表时候oracle内的建表语句:

-- Create table
 create table INTVL_ORDERS
 (
  ORDER_ID  VARCHAR2(10) not null,
  ORDER_DATE DATE default sysdate,
  QTY        INTEGER,
  PAYTERMS  VARCHAR2(10),
  BOOK_ID    NUMBER(6)
 )
 partition by range (ORDER_DATE)
 (
  partition P1 values less than (TO_DATE(' 2014-01-01 00:00:00', 'SYYYY-MM-DD HH24:MI:SS', 'NLS_CALENDAR=GREGORIAN'))
    tablespace USERS
    pctfree 10
    initrans 1
    maxtrans 255
    storage
    (
      initial 64K
      next 1M
      minextents 1
      maxextents unlimited
    ),
  partition SYS_P22 values less than (TO_DATE(' 2014-07-01 00:00:00', 'SYYYY-MM-DD HH24:MI:SS', 'NLS_CALENDAR=GREGORIAN'))
    tablespace USER01
    pctfree 10
    initrans 1
    maxtrans 255
    storage
    (
      initial 64K
      next 1M
      minextents 1
      maxextents unlimited
    )
 );
 -- Create/Recreate primary, unique and foreign key constraints 
 alter table INTVL_ORDERS
  add constraint INO_PK primary key (ORDER_ID)
  using index 
  tablespace USERS
  pctfree 10
  initrans 2
  maxtrans 255
  storage
  (
    initial 64K
    next 1M
    minextents 1
    maxextents unlimited
  );

结果自动扩展了一个分区!这也就是间隔分区的特点.
间隔分区会根据起始区间和一个间隔(之前我们设置的是一个月),插入数据时,
会创建相应的分区,而不会因数据超出分区范围导致语句执行失败.

<引用分区>
 引用分区通过父表继承分区键,可以在逻辑上均分具有分子关系的表.
 分区键通过现有的父子关系解析,由现行的主键或外键约束实施.

SQL> create table range_books
   (bookid number(6) primary key
   ,bookname varchar2(40) not null
   ,booktime date default sysdate
   )
   partition by range(booktime)
   (
   partition p1 values less than (to_date('20140331','yyyymmdd'))
   ,partition p2 values less than (to_date('20140430','yyyymmdd'))
    );
  
 Table created
 

SQL> create table ref_order
   (order_id varchar2(20) constraint RFO_PK primary key
   ,order_date date default sysdate
   ,book_id number(6) not null--not null !!!
   ,constraint RFO_FK foreign key (book_id) references range_books(bookid)
   )
   partition by reference (RFO_FK)
   ;
  
 Table created

上面的实验,通过使用外键发现分区机制,外键[RFO_FK]指向父表'range_books',
因此,子表'ref_order'就按照父表的分区方式进行相应的分区.


注意:创建引用分区表时,通过[partition by reference]子句指定分区方法,
其后括号中指定分区使用的约束.外键约束引用的列必须具有not null约束.
=========================================================================
 分区表维护:
主要的就是 删除分区 | 增加分区 | 合并分区 | 移动分区 | 重命名 | 截断分区

删除分区: 
	alter table range_orders drop partition p2;
	
增加分区: 
	alter table range_orders add partition p2 values 
	less than (to_date('20140430','yyyymmdd')) tablespace user02;
	
合并分区: 
	alter table range_orders merge partition p1,p2 into partition p2;
 
移动分区: 
	alter table range_orders move partition p2 tablespace user05;
 
重命名: 
	alter table range_orders rename partition p2 to p1;
 
截断分区: 
	alter table range_orders truncate partition p1;