It would be helpful if the author included the round-trip latency (even some simple pings) alongside the benchmark so we could judge how far the servers are fro

It would be helpful if the author included the round-trip latency (even some simple pings) alongside the benchmark so we could judge how far the servers are from the client. This will have a significant impact on the ab numbers (particularly as he's not using the -k option and therefore establishing a new TCP session for every request).