Bulk Throughput Measurements - JLab
Bulk Throughput Measurements | Bulk Throughput Simulation | Windows vs. streams | Effect of load on RTT and loss | Bulk file transfer measurements
|
Bulk Throughput Measurements - JLabBulk Throughput Measurements | Bulk Throughput Simulation | Windows vs. streams | Effect of load on RTT and loss | Bulk file transfer measurements |
The measurements were made between pharlap.slac.stanford.edu and hpcdev01.jlab.org
located at the Jefferson Laboratory Norfolk Virginia.
Hpcdev01 was a Pentium 4 1496MHz cou running Linux 2.4.2-2.
It has a 100Mbps Ethernet connection to the JLab network and JLab had
a 155Mbps connetion to ESnet and the Internet.
The pipechar
from SLAC to JLab indicates that the route is via ESnet to Sunnyvale where it
Measurements made on October 3 '01, from pharlap to hpcdev01 are shown to the right.
They indicate that
the throughput maxima are around 58Mbps (the top 10% throughputs measured are above
58Mbps).
We were unable to measure the ping RTT from SLAC to hpcdev01 or www.jlab.org.
We assumed this was due to
security measures that had blocked ping.
So we measured it from SLAC to another host at JLab where we knew ping would respond.
The unloaded RTT from SLAC to www.jlab.org was min/avg/max
113/113/119 msec.
This corresponds to
a window size of about 1.4MBytes for a bottleneck of 100Mbits/s.