Re: Question on V$SESSION_EVENT??

One suggestion I would have is to drop the index before doing the insert and then recreate it afterwards. For every row inserted, the index needs to be updated, and since date indexes are the least efficient of indexes, I'm sure it is slowing you down quite a bit.

ccdicky in hotmail wrote:

> Hi there,
>
> Select 50,000 rows (100 bytes per row) from table A and then insert into
> table B taking about 15mins to get done. However, it takes less than 1min
> if spooling to a file. Even table B contain an index on date field, it
> still doesn't make sence to me. Since the both tables are stored in
> different tablespaces with different disks. In addition, there is no any
> other running process during the insertion.
>
> Therefore, during insertion, I tried to query a number of performance views
> and discovered that the "total_waits" of 'LATCH FREE' event is exactly same
> as the "total timeouts" of that in V$SESSION_EVENT view. Is it the reason
> why the insertion is running so slow? And what is the meaning of this
> event? If not, any suggestion to find out the real cause.
>
> Any pointer would be appreicated. Thanks!
>
> Dicky

Received on Fri Nov 12 1999 - 13:41:30 CST