With Vern Paxson & the people from the Pittsburgh Supercomputer Center
pinned down what we need to order to get started
on NIMI deployment.
Prepared, got review, completed and distributed 30 page
December 1997 Interim Report of the ICFA-NTF
Monitoring Working Group.
Wrote a one page
summary of networking activities for the
ESnet Yearly Report.
Attended the January XIWT/IPWT phone conference.
Worked with Jim Leighton and Chin Guok of ESnet to understand the ESnet monitoring reports
on link utilization that go back to 1990. Re-analyzed the data to provide reports on long term utilization
trends for several HENP sites.
With Valeri Federov, a statistician from ORNL, discussed statistical experiment
design techniques and how they might be relevant to
optimizing the selection of collection sites etc. Also provided Valeri with information on
how to access our raw data.
Worked with Elsevier to push forward publication of work on network monitoring
in Computer Physics Communications.
Consulted with Walter Toki of Colorado State and also a member of the ESSC, on issues
to do with the performance of the SLAC - Colorado State link.
Prepared Network Monitoring - Status
report for the ESSC. Attended ESSC meeting in Washington and presented talk.
Had many useful discussions with attendees.
Grouping the Data
- Tables for metric-by-site are no longer generated from the timeping data
by SAS, so Warren adapted the routine from
group.pl to allow groups to be shown in
- This option is not fully functional yet, but he will make it so the user
can click on group name to view members of the group, e.g.
Long Term Analysis
- Warren has completed the routines to calculate values for MTBF (mean time
between failures) and MTTR (mean time to repair). He's working on a glossary
to define terms and show the mathematics used to calculate the various values.
- A perl script generates reports for
- The perl script generates static html and text versions, the text version
is used to display the results dynamically using
Network Maps / Anemone
- Warren has collected some sample data of traceroutes to each of the nodes we
- The subsequent network map was too complicated to be of use, so he
divided them according to groups;
N America (East),
N Amercia (West),
- I will set up regular tracerouting, to show the current status of the
network on the maps.
Back to Top
- Warren added a few lines to the ping collection program to email a warning
if no packets are received back from the remote site. This is intended to
keep us informed of potential problems, no attempt is made to send automated
email to the contact for the remote site.
Revised 29 January 1998