on 11-25-2010 3:13 PM
Hello all,
run various SAP systems on Oracle 11.2.0.1 (ERP 2004, BI 7.0. PI 7.0,
SRM 4.0, SOLMAN 7.01 )
I have numerous alerts every day as per below, in most of these systems
Redo log buffer [ 10 Alerts ], 2.288 < 4.000 redo entries per redo
log space requests
In
http://help.sap.com/saphelp_nw70ehp1/helpdata/en/45/e2467adee64ebee10000000a11466f/frameset.htm
I read:
Redo log buffer
Cause: The number of redo entries per redo log space request fell below
the configured threshold.
Oracle recommends at least 5000 redo entries per redo log space request.Action: Increase the size of the redo log buffer by increasing the
init<DBSID>.ora parameter
log_buffer until the number of redo entries per redo log space request
stops increasing. .
However, when I upgraded to Oracle 11, in note 1431798 - Oracle
11.2.0 : Database parameter settings ,
it reads that we shouldn't be setting the LOG_BUFFER value and to use
the default. The default set automatically by Oracle in my systems is between 16M and 21M
In addition, I do not understand how increasing the size of the redo log buffer would make the number of redo entries per redo log space request bigger. I understand that a redo log space request is a redo log switch, which will happen as soon as the respective redo log fills up and I do not see how this relates to the size of the redo log buffer itself.
Please let me know what you think
Many thanks
Andreas
Well i am interested in the answer of the "gurus" here
In my opinion the SAP recommendation is totally unclear and your systems seem to be fine. As you said the log_buffer parameter shouldn't be set and sizes around 20mb look perfectly ok to me.
The redo log buffer is flushed
o every 3 seconds
o when 1/3 full
o when you commit.
-> [Ask Tom on redo log buffer size|http://asktom.oracle.com/pls/apex/f?p=100:11:0::::P11_QUESTION_ID:621023586146]
There is no direct relation on the number of redo log entries and the redo log buffer size...
Cheers Michael
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hello Andreas
Increasing the LOG_BUFFER parameter should reduce redo log file IO.
Log switches normally takes place after a redo log file is full but they can also take place at certain intervals. In case of a redo log switch at a certain interval, it would mean that if the IO is reduced, there can be more entries written into the redo log file, thus perhaps hitting the given metric in the monitoring.
If there isn't enough space in the redo buffer, a redo space wait occurs, you can check if it's the case with following SQL statement (result should be < 1)
select round((req.value/wrt.value)*100,2)
from v$sysstat req, v$sysstat wrt
where req.name='redo log space requests' and wrt.name='redo writes'
Kind regards
Tom
User | Count |
---|---|
93 | |
10 | |
10 | |
9 | |
9 | |
7 | |
6 | |
5 | |
5 | |
4 |
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.