Analysis statistics have been gathered for running RHAMT on Openshift using 2 different templates, and on a local Laptop.
The Openshift templates differ from each other in that the shared storage configuration has a shared volume between the Web and Executor Pods which the reports are written to/read from, where as the REST configuration uses http to transfer the Reports from the Executor to the Web Pod.
The results of the analysis are here
The Report Rendering Phase of the analysis process is consistently the longest elapsed time (up to 59% of the total analysis time).

The purpose of this jira is to see what can be done to make that phase more efficient.

One contributory factor may be the number of cores that the application believes to be available. There is a know Java 8 issue whereby the number of cores on the node, rather than the number of cores in the container is returned. This has been fixed in Java 10

The call to determine the number of available threads is here