Bandwidth to the World Project Status 
Les Cottrell, 11/03/01

We now have iperf running on about 23 hosts. The aggregate throughput
from SLAC to these hosts is still about 1.3Gbps. In the last week we obtained accounts 
and machines at FNAL and GSFC and have configured them successfully. Performance to
the FNAL host was improved by a factor of 4 to 5 after it was configured for
100Mbps switched full duplex from half duplex and also a mismatched half/full
duplex configuration Sun is going
to provide access to a SunFire E15K to be located at SC2001. Les tared  many of the
scripts and executables and has emailed them to Rocky Snyder of Sun to install
them on the machine while it is staged in Colorado. We await machines and accounts at
LANL, Michigan and are trying to resolve problems with iperf (maybe ports) and 
ssh at Rice which are preventing us running both iperf and bbcp. 

Andy Hanushevsky has made some improvements to bbcp to get around problems encountered
with hosts with no DNS entry, 64 biyt libraries, and missing compression libraries (libz).

Olivier Martin has provided
a pointer to some code to help with reliability of iperf servers that are supposed
to run all the time, but we have not had a chanvce to implement it yet.
The preferred method of using ssh to automatically
start and stop the servers ran into problems of the Kerberos token needed
to access the ssh keys expired after 25 hours. Connie is looking at
using trscron to get around this, in the meantime we are renewing the
Kerberos token by hand on a daily basis.

ESnet are working on providing a test set up with access to the OC12 
link for SC2001. At the moment we are still running with the OC3 ESnet link.

Steffan and Warren have configured stand alone Linux machines that will be
sent to Denver on Monday 5th November. Connie and Les ported the scripts and 
executables etc., and set ssh keys on these machines so they are now also
successfully using iperf and bbcp to monitor performance. We had to fix an 
ssh version problem and also had problems getting gnuplot to work properly.
Steffan has installed the Apache web server so we can use that to server
up the reports.

Jerrod is starting to work on comparing the bbcp and iperf throughputs, and
Warren is preparing an animated gif with iperf and bbcp throughputs shown on 
a map as a function of time. Connie worked on reading out the switch MIBs
and preparing timeline plots of the throughput and other variables.

The web page for the project is at:
http://www-iepm.slac.stanford.edu/monitoring/bulk/sc2001/