cancel
Showing results for 
Search instead for 
Did you mean: 

Distributed vs. Central System

Former Member
0 Kudos

Hi,

Currently I am in a scenario where my Database (DB2) and application is running on the same central system. I have endusers in almost 100 supply office and almost around 1 million consumer data. The system is the ISU Production server with ERP 6.0 running on AIX 6.1 in IBM P570 server. Recently we are experiencing some performance issues with the system like slow performance being a major issue. For the ISU production server we also have a failover server of same configuration which is at the moment almost unutilised.

Will it be wise to distribute the application and database between the two servers other that running both on the same? If so what will be the best procedure to do this sort of change as we are already in Production environment?

Thanks & Reagrds,

Daipayan Basu

In summary

DB and CI are in the same phisical machine. Can we place DB and CI in different phisical machines? All the mount points of ISU Production server are in SAN storage.

Regards,

Daipayan Basu

Edited by: Daipayan Basu on Feb 13, 2012 6:41 AM

Accepted Solutions (1)

Accepted Solutions (1)

Former Member
0 Kudos

Please, before changing your configuration out of the blue, you should investigate in what is causing your performance problems. Only then you should decide on what actions to take.

DB and CI are in the same phisical machine. Can we place DB and CI in different phisical machines?

Of course this is possible, but this configuration can be slower, because you have a longer way to communicate from DB to CI. Only if your primary server is overloaded, then the distribution will help. But in that case you will still have a problem when one of the nodes fails.

Cheers Michael

Former Member
0 Kudos

Hello Michael,

Thanks a lot for your feedback. I will send you details about some of the performance issues that we are facing in our production environment. But before that I would like to share with you the breif architecture of our production environment.

At present our DB and CI is running on Node1 and there is a failover node (Node2) with same configuration. At the moment this Node2 is acting as standby server. What we are planning is to move tthe DB from Node1 to Node2 (failover server) and configure the cluster in such a way that in case of failure of Node1 CI will failover to Node2 (DB Node) and vise versa.

By doing so we can atleast allocate more RAM and CPU to both DB and CI.

In the present configuration we observe that a few z programs are consuming high memory taking work process in private mode.

Do you think segregration of DB and CI node may help resolve these issues?

I will be eagerly waiting for your response.

Warm Regards,

Daipayan Basu

Former Member
0 Kudos

Do you think segregration of DB and CI node may help resolve these issues?

Short term - probably yes, but what do you do when you have to switch because of a hardware failure? The remaining single node will lack memory and start paging.

I suggest you consider a memory extension with your hardware partner, memory is very cheap today. And i strongly suggest to check those Z-programs, very often they can be optimized easily (not loading all data to an internal table at once etc).

Cheers Michael

Answers (0)