I'm looking to come up with the best combination of Test Packet Size and numbers of Test packets to run to get the closest approximation to 'real world' performance. Any suggestions?

One other...