Right, peak rate doesn't tell the whole story. The description of this
project says:
262 Gbytes of data from LBL to SC2000 show floor
5 sec peak rate of 1.48 Gbps
60 min ave of 582 Mbps
see http://www-didc.lbl.gov/presentations/SC00.LBNL.netchallenge.pdf for
more details.
Rich
At 04:23 PM 2/7/01 -0800, Mahadevan Iyer wrote:
Richard Carlson wrote:
> You're right, the TCP peak was 1.48Gbps from the show floor in Dallas to a
> storage cluster at Berkeley Laboratory. Single applications using multiple
> stream. Bottleneck link was 1.5Gbps provisioned circuit on Qwest link from
> convention center to DARPA's HSCC pop node.
>
> This beat last years (SC'99) 1.2 Gbps rate Microsoft achieved running TTCP
> on a single PC with 2 Gbit Ethernet cards (Redmond to Portland).
>
> Rich
>
Peak rate alone is a meaningless measure of performance. The TCP session may
have reached peak for a small fraction of the session duration and the rest of
the time, the link may have been severely under-utilized.
A meaningful performance measure here is average or percentile link
utilization
over the tcp session duration.
------------------------------------
Richard A. Carlson e-mail:
RACarlson(_at_)anl(_dot_)gov
Network Research Section phone: (630) 252-7289
Argonne National Laboratory fax: (630) 252-4021
9700 Cass Ave. S.
Argonne, IL 60439