on 05-11-2010 9:42 AM
Hi,
We have upgraded our Test environment from MDM 5.5 to 7.1. We are experiencing performance problems, even though the CPU and memory usage in the MDM servers are low.
We get the following message when performing small tasks in Data Manager and Syndicator:
"Wait while the MDM server completes an extended operation"
We have checked the following link.
http://help.sap.com/saphelp_nwmdm71/helpdata/en/45/aef3a652dd52f5e10000000a1553f7/content.htm
Possible reasons we can see:
- many records in repository (800.000+)
- Too many Keyword search enabled fields
- Too many Searchable fields
However, I do not see how these things have such a big impact on the performance. It was a lot better in the 5.5 system, with the same settings in the repository and the same number og records. We are running an Oracle DB.
Is there any settings that may be set in MDS.ini-file?
Is there any other reasons you can think of?
Kind Regards,
Thomas
Hi
After SAP MDM 7.1 upgrade what are functionalities has been changed
- Main table
- Lookups
What is current volume of data ?
Thanks,
Satnam Dhillon
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
I've had exactly the same experience with the 5.5 to 7.1 upgrade - DataManager performance is degraded with no functional changes to the landscape (no additional MDM components, same data volume etc).
(We're still on 7.1 SP02)
Mostly via a process of elimination, we narrowed the problem down to two areas:
- MDSS: Maps take much longer to execute. While these are running, the DataManager displays the wait message
- Java API: When there is i/o in progress via the API, the wait message is displayed
I am working through a MDSS-focused customer message with development right now - a fix for MDSS-performance is apparently included in SP05.
Please PM if you'd like more details.
...Dee
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hi Thomas,
we had simmilar problem after upgrade from 7.1 SP3 to SP4.
We have rised OSS notes to SAP and their info was:
1. Please check note - 1264660 and see if you have the right client
installed.
2. clix mm is a command that is running against MDS. You need to
specify the MDS host name. For example - If you run clix mm from the
mds host machine it should be: clix mm localhost
When we run this clix mm mdm_server_host they have identifyed that the low performance is because could be related to MDSS connecting and trying to get Port and timestamps info from MDS and locking the repository.
When we turn off MDSS the issue dissapera and the performance was fine.
They also ask us to check SAP Note - 1462273 and to increase the parameters to 600:
Auto Syndication Task Delay (seconds)=600
Manual Syndication Task Delay (seconds)=600
When we started the MDSS everything was fine.
So try to increase the syndication delay a little and to turn of the MDSS - hope that it will help.
BR,
ILIN
Hi Thomas,
Please go through the Document release by SAP (MDM 7.1 Good to know topics. You can check some of the MDS.ini settings like "Max threads per Operation" ,Extra DB Validation and Accelerator files and etc. Please recreate the Indexes also and it will help .
Thanks
Ganesh Kotti
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hi Thomas,
I don't think that there is any setting for Performance in Mds.ini file.
I would suggest to compare sizing guide for MDM 5.5 and MDM 7.1 where there is difference in value of Sizing CPU, sizing Memory and Disk sizing. Moreover if you are using Tuple which also one of the significance factor of decreasing performance.
I hope if you carefully compare these sizing guide's, you will able to find the reason for performance difference.
MDM 5.5 Sizing Guide: https://websmp109.sap-ag.de/~sapidb/011000358700006267202006E
MDM 7.1 Sizing Guide: https://websmp109.sap-ag.de/~sapidb/011000358700001921872008E
Regards,
Mandeep Sanii
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Thanks for your reply.
We have followed the sizing guides, and when we have permormance problems we are not using anywhere near available CPU and memory. In other words, there are lots of free resources on the server when we experience performance problems.
We do not use tuples either.
The Repository looks exactly the same as it did before the upgrade, on the 5.5 MDM system, where we did not experience performance issues.
KR,
Thomas
User | Count |
---|---|
84 | |
10 | |
10 | |
10 | |
7 | |
6 | |
6 | |
5 | |
4 | |
4 |
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.