How to calculate server processing time between two UITs

If you have exhausted the options available in portal, you could consider creating a CDT that has a field for each node, update that specific field when the node completes, and write the cdt to a database at the end of the process or before the last UI task.
If the process is likely to change drastically or has a very large number of nodes this becomes less efficient. but would provide the metrics you are looking for and make aggregation very simple.

If you have exhausted the options available in portal, you could consider creating a CDT that has a field for each node, update that specific field when the node completes, and write the cdt to a database at the end of the process or before the last UI task.
If the process is likely to change drastically or has a very large number of nodes this becomes less efficient. but would provide the metrics you are looking for and make aggregation very simple.