In my previous blog post here I described how to get started using Nastel’s free MQSpeedtest application (available here) to check the performance of your MQ environment.
Today I decided to try it out in an architecture with multiple queue managers. I have an application architecture that looks like the following:
So an application starts the transaction by putting to Queue Manager A. This is sent to Queue Manager B. An application receives that message, does some processing and then sends a related message on to Queue Manager C where the final part of the transaction processing occurs.
I wanted to set up some monitoring in parallel to this so that I could see that the MQ flow itself from A to B to C was performing adequately. So I set up the following ping architecture:
I ran the following command:
MQSONAR RTOQMCPCF QMA -fmqsonar.csv -a -d5
Converting the output into a chart looked like this:
Next I ran the MQSpeedtest from A to B (plotted as QMB) and the MQSpeedtest from A to B to C (plotted as QMC) at the same time. This produced the following graph.
So here you can see that:
• I started recording before starting the channels, so initially the route to both QMB and QMC was blocked.
• I started both channels which meant that the route to both QMB and QMC was fine
• I stopped channel QMB.TO.QMC which meant that the route to QMC was blocked but the route to QMB was fine
• I started channel QMB.TO.QMC which meant that the route to both QMB and QMC was fine
• I stopped channel QMA.TO.QMB which meant that the route to both QMB and QMC was blocked
• I stopped channel QMB.TO.QMC which had no immediate effect but affected the following step
• I started channel QMA.TO.QMB which meant that the route to QMB was fine but the one to QMC was still blocked
• I started channel QMB.TO.QMC which meant that both routes were fine again
Following this I can start forwarding the data to Nastel’s AutoPilot and XRay products for real time availability monitoring and historical trend analysis.
The previous blog in this series cal be viewed here
This blog post was originally posted here