Hi all,
We are implementing ESS/MSS on EP 6.0 SP19. Currently we are in the process of doing stress test for the ESS/MSS for 5000 users. We are facing some challenges as described below. I am looking for someone to help from the experienced people in ESS/MSS.
We have Portal running on AIX. We are using 2 Portal Application servers load balanced using CSM (Cisco Switch Module) and running stress test scripts using Mercury Load Runner. Both the Portal Application servers have 4 CPU each.
1. When we run certain scripts ( e.g time statement or pay statement) from mercury, with the multiple iteration ( that means all the users after logging into Portal continue to become active and keep performing certain steps to see their pay statement for different months and all of them are active), users start failing after we reach to the load of around 300 users. At this time none of the Portal Application servers are overloaded and both CPU and memory are well under the limit. When we look at the various error logs in Portal, we do not see any errors. Therefore is is difficult to say why user load is failing even with 300 users.
2. When we run the same script with the single iteration ( that means user after logging into the Portal only runs one step like view one month pay statement and then stays idle), we see that each application server can only take about 1000 user load and then starts failing because CPU is 100% utilized.
My questions related to both scenarios are:
I. Did anyone experience the above issues while performing ESS/MSS stress testing before?
2. Have any used CSM load balancer anywhere in conjunction with ESS/MSS for Portal load balancing?
3. Do we know someone who has experience with webdynpro performance tuning to resolve the load and performance issues with ESS/MSS ( E.g high CPU utilization on Portal application server)?
Your help in this matter is greatly appreciated.
Regards
Sandeep