Oracle FAQ | Your Portal to the Oracle Knowledge Grid |
![]() |
![]() |
Home -> Community -> Mailing Lists -> Oracle-L -> Re: UNDO poll - was undo segments vs. rbs
In a few of my databases I ended up using less (20-25% less) disk space for the
undo tablespace. I started with the same amount of space that was allocated to
the RBU method. After monitoring undo generation and maxquerylen, I found that
the default value of 15 minutes (900 seco) for UNDO_RETENTION was way too large
for our requirements (we don't care about FBQ). Setting it to just 60 seconds
was sufficient.
Also, IMO, Oracle's suggestion to set UNDO_RETENTION to max(v$undostat.maxquerylen) is more of a CYA type. This will tend to increase the size of the undo tablespace. One must monitor undo generation, and also, what tables the long running queries are accessing during the DML operations to deal with ORA-1555, not just blindly follow Oracle's suggestion. You will need more disk space to let Oracle deal with those.
Regards,
-- Archives are at http://www.freelists.org/archives/oracle-l/ FAQ is at http://www.freelists.org/help/fom-serve/cache/1.html -----------------------------------------------------------------Received on Wed May 19 2004 - 08:00:56 CDT
![]() |
![]() |