这是以前做的一个测试,拿出来给大家作为参考吧
测试表:mqqtest
测试表原始数据量:0
插入数据量:1000w
分三种方法插入作为对比参数:
1.create table as select 方式
2.insert into方式
3.insert /*+append*/ into方式
第一种方法由于是一个ddl操作,不需写回滚段,因此耗时在dbwr
第二种方法是一个dml操作,在默认情况下是要写回滚段,因此增加了lgwr的额外开销
第三者方法显示的指定insert操作无需写回滚段,因此性能跟第一种方法一样。
方法一:create table mqqtest
as
select * from mqqstat.t_sub_stat where rownum <= 10000000
///////////////////////////////
Table created.
Elapsed: 00:01:53.79
//////////////////////////////
方法二:
insert into mqqtest(STATDATE, SERVICE_CLASS, SERVICE_ID,
STATION, CITY_CODE, REG_MODE,
UNREG_MODE, CHANNEL_CODE, SUBCHANNEL_CODE,
PAY_WAY, FEATURE_STR, TOTALNUM,
OPENNUM, CLOSENUM, CLOSEIN1DAYNUM,
CLOSEIN3DAYNUM, O1REMAIN3, O2REMAIN3,
O3REMAIN3, THISMONTHOPENCLOSE, HISTORYOPENCLOSE)
select STATDATE, SERVICE_CLASS, SERVICE_ID,
STATION, CITY_CODE, REG_MODE,
UNREG_MODE, CHANNEL_CODE, SUBCHANNEL_CODE,
PAY_WAY, FEATURE_STR, TOTALNUM,
OPENNUM, CLOSENUM, CLOSEIN1DAYNUM,
CLOSEIN3DAYNUM, O1REMAIN3, O2REMAIN3,
O3REMAIN3, THISMONTHOPENCLOSE, HISTORYOPENCLOSE
from mqqstat.t_sub_stat where rownum <= 10000000
10000000 rows created.
Elapsed: 00:08:38.57
方法三:
insert /*+APPEND*/into mqqtest(STATDATE, SERVICE_CLASS, SERVICE_ID,
STATION, CITY_CODE, REG_MODE,
UNREG_MODE, CHANNEL_CODE, SUBCHANNEL_CODE,
PAY_WAY, FEATURE_STR, TOTALNUM,
OPENNUM, CLOSENUM, CLOSEIN1DAYNUM,
CLOSEIN3DAYNUM, O1REMAIN3, O2REMAIN3,
O3REMAIN3, THISMONTHOPENCLOSE, HISTORYOPENCLOSE)
select STATDATE, SERVICE_CLASS, SERVICE_ID,
STATION, CITY_CODE, REG_MODE,
UNREG_MODE, CHANNEL_CODE, SUBCHANNEL_CODE,
PAY_WAY, FEATURE_STR, TOTALNUM,
OPENNUM, CLOSENUM, CLOSEIN1DAYNUM,
CLOSEIN3DAYNUM, O1REMAIN3, O2REMAIN3,
O3REMAIN3, THISMONTHOPENCLOSE, HISTORYOPENCLOSE
from mqqstat.t_sub_stat where rownum <= 10000000
//////////////////////////
10000000 rows created.
Elapsed: 00:01:25.73
/////////////////////////