Ideally you want to establish a baseline to compare your throughput
to the maximum possible throughput.
If you can log into both the sending and receiving host, the best
way to do this is using the Iperf tool.
Here are some good default settings to start with.
start receiver: iperf -s -l 256K -w 4M
start sender: iperf -c hostname -i 5 -t 30 -P 4 -l 256K -w 4M -r
This gives you an upper limit on your bulk file transfer network throughput.
If the iperf throughput is much lower than you expect, first verify that
the host tuning on both the sender and receiver hosts has done properly. If
both hosts are properly configured and performance is still low, there may be
congestion in one of the end-site LANs, or it may be a firewall issue.
This can be tricky to diagnose.
A good place to start tracking down these sorts of problems is to use the
Internet2 Network Diagnostic Tool (NDT). Connect to one of
the NDT servers
that is closest to your location.
Other possible techniques are described in
Internet2
Network Performance Tools Cookbook, or
contact your local network expert, or contact ESnet.
It is also good to know what disk performance you should expect.
It doesn't matter if you have a clean 10Gbps network path you are using slow disks.
Writing to disk is typically slower that reading from disk, so try to measure the write speed on your
receive host.
