RE: measurement

The measurement techniques described thus far focus on performance within
the bounds of a single network. While this is of course a challenge, what
about efforts to measure performance _across_ networks? I'm not talking
about NAP packet loss here, but a true measure of expected customer
satisfaction. Most customers do not particularly care why they can't
download the latest MSIE/Navigator release quickly, and getting traffic to
and through other networks is of equal importance. Network rating systems
are starting to emerge, and I think NANOG should participate in their
evolution...

The measurement techniques described thus far focus on performance within
the bounds of a single network. While this is of course a challenge, what
about efforts to measure performance _across_ networks? I'm not talking
about NAP packet loss here, but a true measure of expected customer
satisfaction.

Expectations of customer satisfaction are aleph nul or likely aleph one.

Measurement of end user delivery are being done rather ad hack (hit the web
site and see how high it bounces) by the folk at Intel, see

    Network Working Group J. Sedayao, Intel Corporation
             C. Bickerstaff, Intel Corporation
    Internet Draft
    Expiration Date: May 1997 November 1996

     Simple End to End Metrics and Methods for Monitoring and Measuring IP
            Provider Performance

The IPPM WG is trying to work upward from a sound theoretical base. See the
other IPPM drafts and mailing list archives.

randy

> The measurement techniques described thus far focus on performance within
> the bounds of a single network. While this is of course a challenge, what
> about efforts to measure performance _across_ networks? I'm not talking
> about NAP packet loss here, but a true measure of expected customer
> satisfaction.

Expectations of customer satisfaction are aleph nul or likely aleph one.

And someone called my name...

Measurement of end user delivery are being done rather ad hack (hit the web
site and see how high it bounces) by the folk at Intel, see

    Network Working Group J. Sedayao, Intel Corporation
             C. Bickerstaff, Intel Corporation
    Internet Draft
    Expiration Date: May 1997 November 1996

     Simple End to End Metrics and Methods for Monitoring and Measuring IP
            Provider Performance

The IPPM WG is trying to work upward from a sound theoretical base. See the
other IPPM drafts and mailing list archives.

May I also suggest people take a look a Treno. If used consistently and
regularly it may give a rought estimate of connection quality for end
users across networks.

randy

Aleph One / aleph1@dfw.net
http://underground.org/
KeyID 1024/948FD6B5
Fingerprint EE C9 E8 AA CB AF 09 61 8C 39 EA 47 A8 6A B8 01

NLANR has a summary of measurement tools here
http://www.nlanr.net/Caidants/meastools.html that has a link to more TReno
information and you can try it out from PSC via this WWW forms interface
http://www.psc.edu/~pscnoc/treno.html

Michael Dillon - Internet & ISP Consulting
Memra Software Inc. - Fax: +1-250-546-3049
http://www.memra.com - E-mail: michael@memra.com