I have recently implemented clustering and load balancing in my test Sonic environment, and have been asked to give proof that the systems are sharing the load of transmissions sent to them. Unfortunately, everything is grouped into the one domain, so what is a sufficient way of seeing which system handled which activity? I know I could use something in the ways of WireShark to monitor acitivity across the "wire", but I'm afraid that there will be so much unneeded noise it won't give the proof that I need. Any recommendations of how I can monitor specific Sonic system activity in a load balanced/clustered environment? Thanks.
You can view how connections are distributed across the brokers from the Sonic Management Console, see the 'Viewing Connections' section in the SonicMQ Configuration and Management Guide. Use metrics such as broker.messages.ReceivedPerSecond and broker.messages.DeliveredPerSecond to monitor activity.
This will show my lack of understanding the full Sonic environment. But I used metrics to view my clustered messaging brokers and noticed that each are receiving messages (proving they are load-balanced). My concern is that I was also using ESB Process Tracking inside my workbench and noticed every line listed the same Host server. Obviously that Host server was the Primary Domain Manager. Was the Primary Domain Manager really handling every single step? What does the Metrics for the other systems really mean, are they actually running the service for the step on that system or is everything running on the Primary Domain Manager and all that is being done for load balancing is the other servers when used are simply moving messages from endpoint to endpoint? I just want to make sure processing of services is actually being shared between system. Thanks.