oracle 数据处理方法

我的测试方法
测试环境t4400+2G+5400转/秒硬务。表大小248MB,1997760条数据。

建立一个基表:
create table a_object as
select object_name,object_type,created,rownum+49944 id  from dba_objects
alter table a_object add primary key (id);
建立一个影响表:
create table b_object as select * from a_Object
1 查询速度比较
a_object
SQL> select count(*) from a_object;
 
  COUNT(*)
----------
    499440 
Executed in 0.219 seconds
SQL> select  count(*) from b_object;
 
  COUNT(*)
----------
    499440
 Executed in 0.766 seconds
这个容易看出,a_object有主键,快,没啥好解释的
SQL> alter system flush buffer_cache;
SQL> alter system flush shared_pool;
 SQL> select count(*) from b_Object a; 
  COUNT(*)
----------
   1997760 
Executed in 4.344 seconds
 
SQL> select /*+parallel(a,2)*/ count(*) from b_object a; 
  COUNT(*)
----------
   1997760 
Executed in 2 seconds
加parallel还是有点用,如果cpu更多,可能会更快。
2 插入
a_Object:insert into a_object
select object_name,object_type,created,rownum+998880 id  from b_object
59秒
b_object:insert into b_object
select object_name,object_type,created,rownum+998880 id  from b_object
25秒
有索引肯定慢些,这个没什么好解释的
insert /*+append*/ into a_object
select object_name,object_type,created,rownum+1997760 id  from b_object
107.531秒
insert /*+parallel(a,2)*/  into a_object a
select object_name,object_type,created,rownum+1997760 id  from b_object
175.172秒,还慢些
insert into a_object a
select object_name,object_type,created,rownum+1997760 id  from b_object
197.125秒,最慢
insert /*+append parallel(a,2)*/into a_object a
select object_name,object_type,created,rownum+1997760 id  from b_object
193.703秒,看样子parallel对表的insert 用处不大,用处最大的是append
alter table a_object  nologging 让表不记日志,再
insert /*+append*/  into b_object
select object_name,object_type,created,rownum+1997760 id  from b_object
57.688秒最快
bulk collect insert into
declare type t_object is table of b_object.object_name%type index by binary_integer;
        type t_type is table of b_object.object_type%type index by binary_integer;
        type t_created is table of b_object.created%type index by binary_integer;
        type t_Id is table of b_object.id%type index by binary_integer;
        objects t_object ;
        types t_type ;
        createds t_created ;
        ids t_id ;
        cursor cur is select * from b_object;
        begin
          open cur;
          loop
            fetch cur bulk collect into objects,types,createds,ids limit 5000;
            for i in 1..ids.count loop
              insert into a_object values (objects(i),types(i),createds(i),ids(i)+1997760);
              end loop;
             exit when cur%notfound;
           end loop;
           close cur;
         end;
468.888秒。这个肯定最慢。
改写
 declare
        type object_type is record
        (
        object_name b_object.object_name%type,
        object_type b_object.object_type%type,
        created     b_object.created%type,
        id          b_object.id%type
        );
        type t_object_type is  table of object_type;    
        objects t_object_type ;
        v_counter   PLS_INTEGER := 0; 
        cursor cur is select object_name,object_type,created,id+1997760 id from b_object;
        begin
          open cur;
          loop
            fetch cur bulk collect into objects limit 10000;
            for i in 1..objects.last loop
             insert into a_object values objects(i);
             end loop;
              v_counter:=v_counter+1;
             exit when cur%notfound;
           end loop;
           dbms_output.put_line('counter is :'||v_counter);
           close cur;
 end;
363秒。还是很慢,用forall改写,
 declare
        type object_type is record
        (
        object_name b_object.object_name%type,
        object_type b_object.object_type%type,
        created     b_object.created%type,
        id          b_object.id%type
        );
        type t_object_type is  table of object_type;    
        objects t_object_type ;
        v_counter   PLS_INTEGER := 0; 
        cursor cur is select object_name,object_type,created,id+1997760 id from b_object;
        begin
          open cur;
          loop
            fetch cur bulk collect into objects limit 10000;
            forall i in 1..objects.last
             insert into a_object values objects(i);
              v_counter:=v_counter+1;
             exit when cur%notfound;
           end loop;
           dbms_output.put_line('counter is :'||v_counter);
           close cur;
 end;
快了不少,131秒。这个速度比较理想。
  declare
        type object_type is record
        (
        object_name b_object.object_name%type,
        object_type b_object.object_type%type,
        created     b_object.created%type,
        id          b_object.id%type
        );
        type t_object_type is  table of object_type;    
        objects t_object_type ;
        v_counter   PLS_INTEGER := 0; 
        cursor cur is select object_name,object_type,created,id+1997760 id from b_object;
        begin
          open cur;
          loop
            fetch cur bulk collect into objects limit 10000;
            exit when objects.count=0;
             v_counter:=v_counter+1;
            forall i in 1..objects.last
             insert into a_object values objects(i);
           end loop;
           dbms_output.put_line('counter is :'||v_counter);
           close cur;
 end;
这个有点离谱,居然变成84.6秒,应当是刚才没刷shared_pool和buffer_cache所致。
 alter system flush shared_pool;
 alter system flush buffer_cache;
注意,以上两条仅限测试环境。
刷新后:158.875秒
。。。。继续更改:
  declare
        type object_type is record
        (
        object_name b_object.object_name%type,
        object_type b_object.object_type%type,
        created     b_object.created%type,
        id          b_object.id%type
        );
        type t_object_type is  table of object_type;    
        objects t_object_type ;
        v_counter   PLS_INTEGER := 0; 
        cursor cur is select object_name,object_type,created,id+1997760 id from b_object;
        begin
          open cur;
          loop
            fetch cur bulk collect into objects limit 10000;
            exit when objects.count=0;
             v_counter:=v_counter+1;
            forall i in 1..objects.last
             insert /* +append */  into a_object values objects(i);
           end loop;
           dbms_output.put_line('counter is :'||v_counter);
           close cur;
 end;
加append;134.672秒。当然这个仅仅是测试,显得有点冗余,但当有条件时,写比较复杂的的转换,又不是对单表操作时尽量用bulk collect  forall吧。
以下开始测删除,从傻瓜式删除开始:
delete from a_Object where id>1997760
--377.14秒
alter table a_object drop constraints SYS_C005373
删除索引后再试,结果惨不忍睹。
建分区表:
create table SCOTT.P_OBJECT
(
  object_name VARCHAR2(128),
  object_type VARCHAR2(19),
  created     DATE,
  id          NUMBER
)
partition by range (id) (
  partition p1 values less than (1997760),
  partition p2 values less than (3995520)
  );
alter table p_object nologging;
insert /*+append*/ into  p_object  select * from a_object;
alter table p_object truncate partition p2;
由于索引失效还需重建索引
 alter index SYS_C005787 rebuild nologging;
insert into 花时248秒,truncate 分区1秒,重建索引15秒。
上一步也可以用再线重定义分区表做。。
顺便 提一下这里的统计信息
begin
 dbms_stats.gather_table_stats('SCOTT','P_OBJECT', cascade => true);
 end;
 用这个收集后不准了。
select num_rows,t.EMPTY_BLOCKS,t.BLOCKS*8/1024,t.LAST_ANALYZED,t.logging from all_tables t where table_name='P_OBJECT' and owner='SCOTT'
select table_name,partition_position,num_rows,blocks*8/1024,sample_size,last_analyzed,high_value from all_tab_partitions where table_owner='SCOTT'
select i.INDEX_NAME,i.table_name,i.num_rows,i.blevel,i.logging,i.DISTINCT_KEYS,i.LAST_ANALYZED from all_indexes i where table_name='P_OBJECT'

posted on 2013-10-18 13:45 snowhill 阅读(173) 评论(0)  编辑 收藏 引用 所属分类: 数据库-oracle


只有注册用户登录后才能发表评论。
网站导航: 博客园   IT新闻   BlogJava   知识库   博问   管理


<2006年12月>
262728293012
3456789
10111213141516
17181920212223
24252627282930
31123456

导航

公告

又一年...........

留言簿(3)

随笔分类(13)

文章分类(131)

文章档案(124)

c++

java

linux

oracle

常用软件

其他

网络配置

系统安全

音乐

搜索

最新评论

阅读排行榜