It is normal for SMON to clean up temporary space for killed sessions. The reason that it is so expensive is that SMON does one small space transactions for every extent, whereas foreground processes would use just one large space transaction to do the same job.

You can stop SMON from doing this job by setting event 10061 (level 1) in its session using the Ixora script post_event.sql. Of course, you'll need to turn the event off again later (level 0) to get the space back.

I am facing a rather strange situation and am baffled by it. I had a long running query on an 8.0.5.2.1 database which I had to terminate after about 24 hours. Unfortunately this query ate up almost 2.5G of sort space (temporary contents tablespace) before it was killed and now SMON is freeing up all this space. The problem is that this activity of SMON is causing it to slow down most sql statements on the database (especially ones that require sorting). I find that the situation can be allevitaed by increasing the sort_area_size and retained size at session level. I am not in a position to shutdown the database and would like to know if there is any workaround that you can think of. I am completely confused and an explanation would be of great help.