on 12-09-2010 6:55 PM
Dear All
For production MDM environment we have installed all mdm servers- MDIS, MDSS, MDS and MDLS on the same physical server.
Will this have any impact on the performance. We are using MDM 7.1 and on HP Unix.
Another question is- If we have multiple repositories Vs single repository with multiple main tables( Vendor, Customer and Materials) why the choice should have performance impacts on the server level.
Thanks-Ravi
Hi,
to have all Servers on one box is is possible. Please consult the MDM Sizing Guide and make sure that you have enough memory and CPU power! I would however chose a repository per data object, i.e. material, vendor, customer, ... so that at the end you have several repositories.
Regards,
Erdal
Edited by: Erdal Simsek on Dec 22, 2010 4:27 PM
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hello Ravi,
Please refer this guide: [http://www.sdn.sap.com/irj/scn/index?rid=/library/uuid/70b0caa1-92e5-2c10-c687-a8d05e778e59]
As per the SAP recommendation, you should not install all MDM servers on same machine.
Also it is better to have separate repositories, instead of having all-in one repository.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
>>MDIS, MDSS, MDS and MDLS on the same physical server
How much of data you would be maintaining in MDM system? For general MDM implementations, you should not have any problem installing all MDM components on same server.
As far as your second question is concerned, the design of repository is based upon your business content.
For example, you would want to store customer and products data in separate rep.
Thanks
Aamir
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Ravi,
I don't mean to muddy the waters here, but performance tuning and system sizing is a slightly more complex exercise than simply determining how many records your repository will contain and how many fields it will have. There are many, many considerations that need to be taken into account.
For example. the MDM server's performance can be dramatically improved or worsened depending on things like sort indices, accelerator files, data types and fill rates, validation logic, remote key generation, number of users, number of Java connections, web service connections, etc. The list goes on and on, and that's not even taking into consideration the hardware (multi-processor, RAM, the physical disk configuration, etc.)
With regards to the Import Server and Syndication Server you have to take into account things like map complexity: are you doing free-form searching to filter records in maps? Are your maps designed for main tables, qualified lookup tables, or reference data? How often do imports / syndications occur? What keys are you matching on when importing? Do you plan on importing by chunks? What is the import volume, etc?
Once again, I don't want to scare you, but I also wanted to bring up a few topics for you to think about. There is a reason why SAP and other vendors charge a fee for doing system sizing.
These are just a few small examples, but the list goes on and on. I hope this helps to get you thinking in the right direction when designing your architecture and system landscape. Good luck!
Edited by: Harrison Holland on Dec 10, 2010 2:34 AM
User | Count |
---|---|
92 | |
11 | |
10 | |
9 | |
9 | |
7 | |
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.