10g怎样为单独一个事务设置较大的UNDO段

Yong Huang

Resource plan is not what he wants. Read the second paragraph of msg #6. He wants a huge undo tablespace for that job. After the job is finished, he wants the undo tablespace as small as the business only needs it to be. With a resource plan, you would have a huge undo tablespace all the time, allow the job to use most or all of it and allow other sessions to use just enough. But you still have a huge undo taking lots of storage all the time, causing a long RMAN backup among other things.

To prevent the job from abort on running out of space, alter session enable resumable at the beginning of the job, and make sure you check the job once every 2 hours or more often or make the resumable timeout longer.

fly518

用10g的资源计划管理，就可以满足这个需求，你可以仔细看看资料，ocm考试都有这些内容的

fly518

可以根据资源消费者，设置不同的undo使用量

sqysl

1、这种情况应该是大job耗光了undo空间;
2、既然undo空间不够，就算10g支持你分配单独的undo段，且可以指定不覆盖undo信息，总的undo空间就更不够的；还有，报出ora-01555的事务，未必是这个大job，要知道系统上不只跑着这一个job；
2、undo_retention起作用是有条件的，很多情况下，只增加它没用；

oracle_lch

感谢大家的关注，问题解决了，从应用角度解决的，切小了事物降低了undo的负担。