Does anyone have enabled New Relic monitoring for SPS?
We are seeing behavior where after reaching 40 Requests/Sec SPS, requests slows down for SAML flow where we are using SPS servers to generate SAML. We are trying to figure out why the requests are slowing down when we are putting load the SPS servers.
Also, when we try to increase the JAVA heap size on SPS the SPS services are not coming up? We are running SPS on windows 2008 R2 servers and SPS version is R12.52Sp1 Can somebody please suggest how to:
1) Get New Relic working with SPS
2) Increase requests processing on load like 40 requests/Sec
3) increase JAVA Heap size
I am not sure on the New Relic setup for SPS monitoring. However Jason_Wilcox has narrated wonderfully about SPS/AccessGateWay monitoring in the below article:
CA Access Gateway Performance Monitoring: Even Simpler Than 1-2-3
Especially focus on Jason's response for one of the query in the comments section:
"That being said, the integration with APM is simply posting a JSON message to a listener on the EPA Agent. If you have another source that could accept the post and parse the JSON body you could still extract that data and utilize it."
Also have a look at these tuning guides:
SPS Tuning on Linux - CA Knowledge
On your point  about increase JAVA Heap size. Refer to the below blog.
SPS CPU utilization 100%
You are running CA AG on windows 2008 R2 servers and CA AG version is R12.52Sp1, which is a 32bit product, hence irrespective of the strength of your server, you'll always be restricted to 1.5 GB of JVM Heap Size. Anything beyond, the CA AG will not start up.
You have two options.
A. Build multiple instances of CA AG to suffice your requirements.
B. Upgrade to R12.8 which is now a 64bit CA AG, thus you can fully explore the potential of tuning JVM Heap Size.
HubertDennis thanks for your response, but I tried to increase the Java heap size to 1024 by default it's 512, but after making that changes also sps service didn't start.
As I said, The problem we are trying to solve is we have so much resources available on the server but it's not fully utilized. We have 8gb ram and 4 could we are expecting to hit a point where we use all the resources and then services/requests slows down but we know we are hitting just 40 requests/sec and requests are slowing down.
If we consider configuring additional instaces then we would need to configure all the instances to hit saml service and hit all of them at the same time, but in production doing that will not be possible. I am not sure if we will be able to upgrade SPS without upgrading policy servers.
Yes, if you upgrade to SPS /AG12.8 you will have to upgrade the policy servers to 12.8.