Oracle 分区,索引,测试 (1)

 -----本次只做插入.看看怎么插入速度快

 

 

准备工作

--数据文件
alter system set db_create_file_dest='d:\toby\oracle\data';
--表空间
create tablespace ts_sales_200901 datafile size 5M autoextend on ;
create tablespace ts_sales_200902 datafile size 5M autoextend on ;
create tablespace ts_sales_200903 datafile size 5M autoextend on ;
create tablespace ts_sales_200904 datafile size 5M autoextend on ;
create tablespace ts_sales_200905 datafile size 5M autoextend on ;
create tablespace ts_sales_200906 datafile size 5M autoextend on ;
create tablespace ts_sales_200907 datafile size 5M autoextend on ;
create tablespace ts_sales_200908 datafile size 5M autoextend on ;
create tablespace ts_sales_200909 datafile size 5M autoextend on ;
create tablespace ts_sales_200910 datafile size 5M autoextend on ;
create tablespace ts_sales_200911 datafile size 5M autoextend on ;
create tablespace ts_sales_200912 datafile size 5M autoextend on ;
create tablespace ts_sales_201001 datafile size 5M autoextend on ;
create tablespace ts_sales_201002 datafile size 5M autoextend on ;


create table city(
city_id number(10),
city		nvarchar2(30),
primary key(city_id)
);

create table employee(
EMPLOYEE_ID number(10),
FIRST_NAME		nvarchar2(30),
LAST_NAME		nvarchar2(30),
MANAGER_ID number(10),
primary key(EMPLOYEE_ID)
);

--- insert city  
--id 从1到24
INSERT INTO CITY
SELECT ROWNUM,CITY FROM HR.LOCATIONS;

-- insert employee
--id 从100到206
insert into employee
select   EMPLOYEE_ID,FIRST_NAME,LAST_NAME,MANAGER_ID  FROM HR.EMPLOYEES;


---分区表
create table sales_data(
sales_date date,
city_id		number(10),
employee_id	number(10),
sales_type  nvarchar2(30), 
sales_amount number(10)
)partition by range (sales_date)
(
partition sales_200901 values less than (to_date('2009-02-01','YYYY-MM-DD')) tablespace ts_sales_200901,
partition sales_200902 values less than (to_date('2009-03-01','YYYY-MM-DD')) tablespace ts_sales_200902,
partition sales_200903 values less than (to_date('2009-04-01','YYYY-MM-DD')) tablespace ts_sales_200903,
partition sales_200904 values less than (to_date('2009-05-01','YYYY-MM-DD')) tablespace ts_sales_200904,
partition sales_200905 values less than (to_date('2009-06-01','YYYY-MM-DD')) tablespace ts_sales_200905,
partition sales_200906 values less than (to_date('2009-07-01','YYYY-MM-DD')) tablespace ts_sales_200906,
partition sales_200907 values less than (to_date('2009-08-01','YYYY-MM-DD')) tablespace ts_sales_200907,
partition sales_200908 values less than (to_date('2009-09-01','YYYY-MM-DD')) tablespace ts_sales_200908,
partition sales_200909 values less than (to_date('2009-10-01','YYYY-MM-DD')) tablespace ts_sales_200909,
partition sales_200910 values less than (to_date('2009-11-01','YYYY-MM-DD')) tablespace ts_sales_200910,
partition sales_200911 values less than (to_date('2009-12-01','YYYY-MM-DD')) tablespace ts_sales_200911,
partition sales_200912 values less than (to_date('2010-01-01','YYYY-MM-DD')) tablespace ts_sales_200912,
partition sales_201001 values less than (to_date('2010-02-01','YYYY-MM-DD')) tablespace ts_sales_201001,
partition sales_201002 values less than (to_date('2010-03-01','YYYY-MM-DD')) tablespace ts_sales_201002
);


--分区索引放在对应表空间
create index index_sales_data_partition on sales_data (sales_date) local 
(
partition sales_200901 tablespace ts_sales_200901,
partition sales_200902 tablespace ts_sales_200902,
partition sales_200903 tablespace ts_sales_200903,
partition sales_200904 tablespace ts_sales_200904,
partition sales_200905 tablespace ts_sales_200905,
partition sales_200906 tablespace ts_sales_200906,
partition sales_200907 tablespace ts_sales_200907,
partition sales_200908 tablespace ts_sales_200908,
partition sales_200909 tablespace ts_sales_200909,
partition sales_200910 tablespace ts_sales_200910,
partition sales_200911 tablespace ts_sales_200911,
partition sales_200912 tablespace ts_sales_200912,
partition sales_201001 tablespace ts_sales_201001,
partition sales_201002 tablespace ts_sales_201002
);
---位图
create bitmap index index_sales_data_sales_type on sales_data (sales_type) local ;
--位图连接
create bitmap index index_sales_data_city on sales_data (city.city_id) 
from sales_data,city 
where sales_data.city_id=city.city_id
local ;


---分区表  但不建位图
create table sales_data1(
sales_date date,
city_id		number(10),
employee_id	number(10),
sales_type  nvarchar2(30), 
sales_amount number(10)
)partition by range (sales_date)
(
partition sales_200901 values less than (to_date('2009-02-01','YYYY-MM-DD')) tablespace ts_sales_200901,
partition sales_200902 values less than (to_date('2009-03-01','YYYY-MM-DD')) tablespace ts_sales_200902,
partition sales_200903 values less than (to_date('2009-04-01','YYYY-MM-DD')) tablespace ts_sales_200903,
partition sales_200904 values less than (to_date('2009-05-01','YYYY-MM-DD')) tablespace ts_sales_200904,
partition sales_200905 values less than (to_date('2009-06-01','YYYY-MM-DD')) tablespace ts_sales_200905,
partition sales_200906 values less than (to_date('2009-07-01','YYYY-MM-DD')) tablespace ts_sales_200906,
partition sales_200907 values less than (to_date('2009-08-01','YYYY-MM-DD')) tablespace ts_sales_200907,
partition sales_200908 values less than (to_date('2009-09-01','YYYY-MM-DD')) tablespace ts_sales_200908,
partition sales_200909 values less than (to_date('2009-10-01','YYYY-MM-DD')) tablespace ts_sales_200909,
partition sales_200910 values less than (to_date('2009-11-01','YYYY-MM-DD')) tablespace ts_sales_200910,
partition sales_200911 values less than (to_date('2009-12-01','YYYY-MM-DD')) tablespace ts_sales_200911,
partition sales_200912 values less than (to_date('2010-01-01','YYYY-MM-DD')) tablespace ts_sales_200912,
partition sales_201001 values less than (to_date('2010-02-01','YYYY-MM-DD')) tablespace ts_sales_201001,
partition sales_201002 values less than (to_date('2010-03-01','YYYY-MM-DD')) tablespace ts_sales_201002
);


--分区索引放在对应表空间
create index index_sales_data_partition_1 on sales_data1 (sales_date) local 
(
partition sales_200901 tablespace ts_sales_200901,
partition sales_200902 tablespace ts_sales_200902,
partition sales_200903 tablespace ts_sales_200903,
partition sales_200904 tablespace ts_sales_200904,
partition sales_200905 tablespace ts_sales_200905,
partition sales_200906 tablespace ts_sales_200906,
partition sales_200907 tablespace ts_sales_200907,
partition sales_200908 tablespace ts_sales_200908,
partition sales_200909 tablespace ts_sales_200909,
partition sales_200910 tablespace ts_sales_200910,
partition sales_200911 tablespace ts_sales_200911,
partition sales_200912 tablespace ts_sales_200912,
partition sales_201001 tablespace ts_sales_201001,
partition sales_201002 tablespace ts_sales_201002
);

--一模一样的不分区表
create table sales_data2(
sales_date date,
city_id		number(10),
employee_id	number(10),
sales_type  nvarchar2(30), 
sales_amount number(10)
);

--一模一样的不分区表2
create table sales_data3(
sales_date date,
city_id		number(10),
employee_id	number(10),
sales_type  nvarchar2(30), 
sales_amount number(10)
);

set serveroutput on
---插入数据 1千万
  DECLARE
		
        MAXRECORDS CONSTANT INT:=10000000;
        sales_date int:=0 ;
		sales number:=0;
        I INT :=1;
        city_id int:=0 ;
        employee_id int:=0;
		sales_date1 nvarchar2(20):='';
    sales_month int:=0 ;
    begin
        FOR I IN 1..MAXRECORDS LOOP
		CITY_ID:= ABS(MOD(DBMS_RANDOM.RANDOM,24));
		employee_id:=ABS(MOD(DBMS_RANDOM.RANDOM,106))+100;
		SALES_DATE:=ABS(MOD(DBMS_RANDOM.RANDOM,28));
        sales_month :=ABS(MOD(DBMS_RANDOM.RANDOM,5));
         
	--保证不为0
		while city_id=0 or SALES_DATE=0 or sales_month=0 loop 
        CITY_ID:= ABS(MOD(DBMS_RANDOM.RANDOM,24));
		sales_month :=ABS(MOD(DBMS_RANDOM.RANDOM,5));
        SALES_DATE:=ABS(MOD(DBMS_RANDOM.RANDOM,28));
        end loop;
        
		sales:=ABS(MOD(DBMS_RANDOM.RANDOM,100000));
		sales_date1:='2009-0'||to_char(sales_month)||'-'||to_char(sales_date);
    
		INSERT INTO SALES_DATA3 VALUES
		(to_date(sales_date1,'YYYY-MM-DD'),city_id,employee_id,'toby',sales);
			
        end loop;
    dbms_output.put_line('done!');
    commit;
    end;

 

现在 SALES_DATA3 有1千万数据,分布在1到5月之间.

三个表 

SALES_DATA     分区, 有位图

SALES_DATA1   分区, 无位图

SALES_DATA2   无分区

 

 

一共跑3次 减少偶然性 取最小时间

 

先做 普通的插入

--用时 167.218,137.045,135.247  最小135.247 秒
insert into SALES_DATA1
select * from SALES_DATA3 where sales_date<to_date ('2009-02-1' ,'yyyy-mm-dd');

  再用 nologging 试试

 

--用时 203.28,189.064,167.993	最小167.993 秒
alter table SALES_DATA1 nologging;
insert into SALES_DATA1
select * from SALES_DATA3 where sales_date<to_date ('2009-02-1' ,'yyyy-mm-dd');
alter table SALES_DATA1 logging;

 时间竟然比 logging 还多. .. 理论上应该是短的.  反正至少看起来 nolog 不明显

 

 再试试 append

 

--用时 81.867 ,	89.569,84.574  最小	81.867 秒
insert /*+append*/ into SALES_DATA1
select * from SALES_DATA3 where sales_date<to_date ('2009-02-1' ,'yyyy-mm-dd');

 再试试别的 append + nologging

 

 

--用时 68.559,84.127,84.907    最小68.559 秒
insert /*+append*/ into SALES_DATA1 nologging
select * from SALES_DATA3 where sales_date<to_date ('2009-02-1' ,'yyyy-mm-dd') ;

 似乎 append + nologging 中 nologging 才有点作用.

 

 

电脑垃圾 并行就不测了 粘下语句

alter table SALES_DATA3 parallel;
alter table SALES_DATA1 parallel;
alter table SALES_DATA1 nologging;
insert /*+append*/ into SALES_DATA1 
select /*+ parallel(t,2) */ * from SALES_DATA3 t where sales_date<to_date ('2009-02-1' ,'yyyy-mm-dd') ;

同样位图 对于insert 影响很大 时间可以相差几倍 

alter index index_sales_data_partition modify partition sales_200903 unusable;
alter index INDEX_SALES_DATA_CITY modify partition sales_200903 unusable;
alter index INDEX_SALES_DATA_SALES_TYPE modify partition sales_200903 unusable;
    
-----insert 语句 
-----重建也用并行
alter index index_sales_data_partition rebuild partition sales_200903 NOLOGGING PARALLEL;
alter index INDEX_SALES_DATA_CITY rebuild partition sales_200903 NOLOGGING PARALLEL;
alter index INDEX_SALES_DATA_SALES_TYPE rebuild partition sales_200903 NOLOGGING PARALLEL;