Work related to Collaborations
We have made contact with several of the major HENP collaborations (RHIC, CDF,
D0) representatives to set up some web sites on PingER monitoring targeted at
each collaboration. These can be viewed from the IEPM home page.
Warren worked with PNL to get them added as another PingER monitoring site.
Warren worked with Milt Mallory on the Stanford campus to set up a
PingER site to evaluate the performance differences over ESnet
compared to Internet2. See
Ping Results for SLAC and Campus to BaBar-collaborating Internet2 Universities
for some preliminary results.
To understand some of the results from UT Dallas Warren talked to the THEnet
We received the NLANR-AMP machine (called Halimede). It is installed, configured
and taking data. See NLANR RTT
Les attended the NLANR organized "Measurement and Collaborations Workshop" and
the "Challeneges and opportunities for measurement in a high performance
environment" workshop. Prepared and presented talks on
and Analysis Activities, ESnet and HENP viewpoint, and
End-to-end Monitoring in ESnet/HENP
and the relevance of ping.
Working with Matt Zerkauskas of the Surveyor project, Warren copied Surveyor
data for November 1998 thru May 1999 for paths between SLAC,
CMU, FNAL and CERN.
Warren then reformatted the data into PingER format, so it can be viewed with
pingER tools, see
PingER Site-by-Month History Table for Surveyor data.
Les worked with Rebecca Moss (a new student) and Warren to analyze the
correlation between PingER and Surveyor and Ping and Surveyor (see
Comparison of Surveyor and Ping). Also looked at the sensitivity of the
data to shifts in time stamps between Surveyor and PingER.
RIPE machine (aka Nemertes) was delayed by the US Customs.
Eventually the SLAC people negotiated it's freedom.
Dave Price and Ken Hooker are working on the GPS requirements.
Heard from Bjorn Frogner (CEO Netpredict) that they have been awarded
the phase II SBIR that we assisted them on.
Unfortunately, the student who we hired to work with Bjorn on
this, has had to return to India to address an emergency. Starting to look for
a new student.
Tools & Analysis
PingER and Ping Validation
Les gathered some high statistics ping data on the LAN between machines on the
same segment, machines separated by a switch, machines separated by a router
etc. to understand the impact of switches, routers, line speeds and hosts
on the round trip time (RTT). See
High Statistics ping results for more information.
Warren has been been busy doing the usual chasing up when data is
missing. This month we've had black spots from DESY, DOE
and RAL, with glitches from CMU, NBI and KEK.
Currently the only outstanding issu is with RAL, and Warren is working
with the contact to resolve it.
Worked with John MacAllister of Oxford to get traceping at SLAC to
monitor all the beacon sites, ran into a problem with insufficient resources to
run the program. Ordered more memory for the machine. Documented
the Current Interest in
Traceping. John MacAllister modified traceping to make it more economic to
Kishan has been doing some baselines, looking at performance without
anything on the line. We worked with Becca
Nitzan from ESnet to put load on network. However problems with the
WFQ code have made progress difficult.
Warren upgraded our would-be IPv6 node (Pontus) to
Red Hat 6.0, and he's working on tunneling IPv6 over IPv4 to ESnet.
Interviewed, hired and got Rebecca Moss started working on preparing an
IEPM web site, see
Internet End-to-end Performance
We interviewed and hired another student, Yvonne Gbalazeh, to write
Java code to enquire to an Oracle database andcreate a GUI front end to it.
For more on this project see
Work Related to Publicity and Information
Worked on paper for BaBar. Graphs etc.
Warren registered for INET99, where we intended to present our
Prepared HTML version for conference CD. However, due to a birth and wedding
neither Les nor Warren could attend the INET99.
Les updated the Internet monitoring tutorial with a link to the comparison
of Surveyor and PingER and also to rationalize why we measure 100 and 1000
byte pings (following a question from a user).
Back to Top
Revised 2 July 1999