大量数据快速插入方法探究
快速插入千万级别的数据,无非就是nologging+parallel+append。
1 环境搭建
构建一个千万级别的源表,向一个空表insert操作。
参考指标:insert动作完成的实际时间。
SQL> drop table test_emp cascadeconstraints purge;
Table dropped.
SQL> create table test_emp as select *from emp;
Table created.
SQL> begin
2 for i in 1..10 loop
3 insert into test_emp select *from test_emp; --批量dml,建议forall
4 end loop;
5 end;
6 /
PL/SQL procedure successfully completed.
SQL> select count(*) from test_emp;
COUNT(*)
----------
14336
SQL> begin
2 for i in 1..10 loop
3 insert into test_emp select *from test_emp;
4 end loop
5 ;
6 end;
7 /
PL/SQL procedure successfully completed.
SQL> select count(*) from test_emp;
COUNT(*)
----------
14680064 --1.5千万级别
2 only append
SQL> set timing on
SQL> show timing
timing ON
SQL> insert /*+ append */ into test_goalselect * from test_emp;
14680064 rows created.
Elapsed: 00:00:20.72
没有关闭日志,所以时间是最长的。
3 append+nologging
SQL> truncate table test_goal;
Table truncated.
Elapsed: 00:00:00.11
SQL> insert /*+ append */ into test_goalselect * from test_emp nologging;
14680064 rows created.
Elapsed: 00:00:04.82
发现日志对插入的影响很大,加nologging时间明显大幅缩短;当然这个表没有索引、约束等,这里暂不考究。
4 append+nologging+parallel
SQL> truncate table test_goal;
Table truncated.
Elapsed: 00:00:00.09
SQL> insert /*+ parallel(2) append */into test_goal select * from test_emp nologging;
14680064 rows created.
Elapsed: 00:00:02.86
这里在3的基础上加上并行,性能基本达到极限,1.5千万数据插入时间控制在3S左右。并行在服务器性能支持的情况下,可以加大并行参数。
转载于:https://blog.51cto.com/90sirdb/1794367