As a baseline, we want a test to measure the raw transfer throughput between nodes in the experiment. This will serve as the absolute maximum our system can achieve if there was no overhead. Our system will perform well below the observations measured here. Having this test will document network conditions at the time the test was run as well as providing context for the future experiments. Steps:

• Select K pairs of nodes (A, B) in the cluster.
• With a 1GB file, transfer from A to B and B to A for each pair
• Measure the throughput each way.
• Report the results in a text file (csv, yaml, or json)