on 03-13-2009 2:19 PM
Subject: ECD Database Contention (Waits)
In looking at the database these two SQL statements are causing a large amount of contention (waits and slow performance) u2013 sessions waiting on the update. This is typically an application issue, that you have one session updating a record that other sessions are waiting to also update the same record. Between these 2 SQL statements looks like this accounts for about 75% of the database time waiting.
SELECT "LOCKVALUE" FROM "VIRSA_CC_LOCKTABLE" WHERE "LOCKKEY" = '1001' FOR UPDATE;
SELECT "LOCKVALUE" FROM "VIRSA_CC_LOCKTABLE" WHERE "LOCKKEY" = '1002' FOR UPDATE;
Can anyone offer an idea as to why, when we run a Full Batch Risk Analysis User Analysis, this is slowing our performance and causing the job to run extremely long?
Greg-
This means that the webservice and background threads are being cleaned up, and does not happen on a consistent basis. You probably scheduled the job when this was taking place. Please run the full batch again, and check the performance.
Ankur
SAP GRC RIG
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hi Gregory,
The lock will always happen, so that during the delete / insertion process, no other user does the same.
But that should not create performance issues in 5.2.
Thanks
Prem
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
An OSS note was created and the following metrics were deemed acceptable for the amount of our analysis:
u2022 Generate Rules u2013 17 hours and 14 minutes
u2022 Full User Role Profile ZEQ1 u2013 26 Seconds
u2022 Full BRA User Analysis u2013 5 minutes (compared to 12 minutes previously)
u2022 Full BRA Role Analysis u2013 1 hour, 30 minutes (compared to 1 hour, 51 minutes previously)
u2022 Full BRA Profile Analysis u2013 1 hour, 14 minutes (compared to 1 hour, 33 minutes)
u2022 Management Report u2013 2 minutes (compared to 4 hours, 6 minutes previously)
DIFFERENT TESTS
u2022 Full BRA User Analysis u2013 1 day, 19 hours, 12 minutes (ALL Users)
u2022 Management Report u2013 1 hour, 2 minutes
u2022 Full BRA Profile Analysis u2013 3 hours, 1 minute
u2022 Management Report u2013 1 hour, 5 minutes
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
The only issue revolves around our DB and how it is processing users while performing a Full Batch Risk Analysis User Analysis.
Both of these notes have already been applied.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Reply by SAP
Can you kindly check and implement SAP Note 1044174. Before implementing this one, kindly kill the long running job. After implementing this Note, kindly schedule new job and check how it fares.
Thanks & Regards,
SAP Governance Risk & Compliance
-
We've already had this note in place.....I will keep you posted.
Greg
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
The job (Full Batch Risk Analysis User Analysis) has been running for 17 hours now. It is processing users, but very slowly. Our DBA noticed that the record is being locked for each single process while the other processes wait. I just created an OSS to see what can be done.
Thanks, Ankur!
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.