Auckland-VI: First synchronized multipoint IP header trace

New Message Reply About this list Date view Thread view Subject view Author view

From: Joerg Micheel (joerg@nlanr.net)
Date: Mon Sep 24 2001 - 00:30:49 PDT


The WAND research group has completed the first GPS-synchronized multipoint
IP header trace at the University of Auckland access link. Instrumented were
a total of three different points: the previous ATM-level tap towards Clear
Communications (Auckland's ISP), the external Ethernet hub and the internal
Ethernet hub (separated from each other by a firewall). The measurements
document very precisely the delay and loss as experienced by packets passing
into and out of the university. The queueing delay caused by the bottleneck
access link is an interesting object for studies.

        http://wand.cs.waikato.ac.nz/wand/wits/auck/6/auckland_infra.png

A total of 12 graphs per data file are currently displayed, documenting the
total delay as well as the processing delay (defined as total delay minus the
serialization delay for the packet at the given link layer) between any of the
three measurement points. Unmatched packets are quantified in a table. We are
working on displaying packet loss.

The data set consists of two different measurements. The first part was taken
for one hour with the Ethernet hubs consisting of 3Com 10BaseT boxes. The second,
long trace, for 4 days with 10/100 autosensing CentreCOM hubs. The effect of
the different link settings on the performance can thus be studied.

        http://wand.cs.waikato.ac.nz/wand/wits/auck/6/

The trace data set and the graphs displayed are the work of Klaus Mochalski
of the WAND research group. The data set is also subject to detailed studies
for the forthcoming PhD thesis work by Stephen Donnelly at Waikato. For more
information please contact the authors.

The data set will be made available to researchers via the NLANR PMA project
at the new trace data server:

        http://pma.nlanr.net/
        ftp://pma.nlanr.net/traces/long

As an interim solution we accept requests for samples which will be distributed
by tape. We support DDS-2, DDS-3 and DDS-4 media as well as EXABYTE.

----- Forwarded message from Klaus Mochalski <klaus@cs.waikato.ac.nz> -----

In June a 4 1/2 days trace has been taken at three points in the
University of Auckland Internet infrastructure. Besides the usual
graphs (utilization and flows) we produced several graphs
illustrating the packet delay between the measurement points.

http://wand.cs.waikato.ac.nz/wand/wits/auck/6/

We'll also add the artefacts graphs as soon as possible.

Comments welcome!

----- End forwarded message -----

        Joerg

-- 
Joerg B. Micheel			Email: <joerg@cs.waikato.ac.nz>
WAND and NLANR MOAT			Email: <joerg@nlanr.net>
The University of Waikato, CompScience	Phone: +64 7 8384794
Private Bag 3105			Fax:   +64 7 8585095
Hamilton, New Zealand			Plan:  PMA, TINE and the DAG's


New Message Reply About this list Date view Thread view Subject view Author view

This archive was generated by hypermail 2b30 : Thu Sep 27 2001 - 16:24:41 PDT