What is the best way to determine how much time for an EMR job is spent on Map vs Reduce Tasks?

I am running a custom jar hadoop job in Amazon's AWS EMR, and I want to gather data on how much time is spent running all Map tasks vs time spent running Reduce tasks. Is there a way in the framework to mine this data that I have not found? If not does anyone have any suggestions on the best way to generate this data?

Thank you,