How do your flows compare? - the CPU and elapsed time story

Doina Klinger
Published on July 18, 2013 / Updated on January 30, 2018

In a previous post, we have seen how you can start using this exciting new feature of IBM Integration Bus version 9.0, the capability to compare the live performance data of running flows.

Here we'll look into more details at this function.

You can use the flow comparison view to see if a message flow is getting more than its fair share of CPU. You'll need to click on Average CPU time and the flow that takes the longest average CPU time per invocation will be at the top. Once again, the data and the flow order are automatically updated every 20 seconds.

You can click on the average elapsed time and see the flows that are taking the most, on average, per invocation. The elapsed time is measured from the moment the flow is invoked until the moment it has completed. High average elapsed time, especially when combined with a lower average CPU, is often a sign of a message flow is waiting for an I/O resource. Could this be the case for your flow?

Select a flow and see the how its (highlighted) nodes are doing. You can order the bottom table too and you see the statistics for all the nodes of all the flows that are running, have statistics collection enabled and are deployed on your selected resource.

You can see the node that takes most CPU or elapsed time, on average or in total, or the node that it is most invoked and could be a bottleneck. The node type is included here to give you the first hint of the reason behind the statistics data. Using the IIB's docs for improving performance of specific node types (Compute. Java Compute, XSLT etc) you can try some optimizations. If you have modified your flow and have redeployed, the statistics tables will give you immediate feedback if the changes have improved the performance of your flow.

This cross-flow view of the nodes is also updated and resorted every 20 seconds.

Once you have seen and understood the relative performance of your flows, you might want to focus on a particular flow. This is where the magnifying glass icon, next to the flow name, comes in handy.

Click on it and a new browser tab is opened and a detailed view of the statistics associated with your chosen flow is not available.

To be continued ...