North American Network Operators Group

Date Prev | Date Next | Date Index | Thread Index | Author Index | Historical

Statistics and network analysis workshop

  • From: Hans-Werner Braun
  • Date: Mon Oct 30 17:07:50 1995


Lately we have been discussing with the NSF holding a workshop on
statistics collection and network analysis. We are thinking about 4/5
January at SDSC in San Diego, but that is still open to changes. We
believe we can comfortably accomodate 30 attendees and have an initial
(not yet mailed out) invitation list, but are interested in suggestions
from NANOG and IEPG relative to attendees and contents and
expectations. If we end up with more than 30 people we will likely have
to go by who has expressed most interest by means of position papers or
so. I will attach a draft of the invitation letter we are planning to
mail out.  Please email me with your suggestions (no need to litter on
the mailing lists).


Dear colleague:
The existence of the NSFNET as a "central network" for the research and
education community facilitated research into aspects of aggregate
network traffic patterns and the anomalies in those patterns caused by
the introduction of new or unique applications.  Decommissioning the
centralized NSFNET backbone resulted in an environment with no
dependable public source of statistics on Internet workloads.  Gaining
insight into the nature of current workloads and their resource
requirements, as well as how they change over time, is essential to
supporting Internet evolution.

For example, the disparity in workload profiles within the current
cross section of Internet applications, ranging from high
bandwidth-duration multimedia to historically less demanding e-mail or
remote login, will necessitate revised metrics for describing aggregate
network behavior.  Simple mean or peak utilization figures will be
dangerously ineffective in addressing a service provider's needs,
without also knowing the `transaction' profile constituting and perhaps
dominating those figures.  (See for examples
of traffic flow profiling, including data from the FIX-West multiagency
network interconnection facility.)

The NSF supported Routing Arbiter provides another source of
operational statistics collection, via route servers at the NSF
supported NAPs.  The RA maintains NAP statistics on packet loss and
latency among routing peers, BGP routing protocol message counts, route
server memory utilization, and selected interface counts on the
server.  Details and sample statistics on the MAE-East facility are at

To facilitate discussion among communities of academia, equipment
vendors, and service providers, who share interest in and incentive to
understand one another's views on network statistics and analysis, we
will hold a workshop with support of the National Science Foundation.
In this workshop we expect to identify, elicit and evaluate the cross
sections of interest, goals, willingness, and technical capability to
facilitate such studies in the new distributed environment.  Critical
to the discussions will be the presence of those who can set policy for
design and configuration of Internet components: vendors who can design
statistics collection mechanisms, and service providers who can ensure
the effective employment of those mechanisms to support their short and
medium term engineering requirements to deliver and maintain a stable
on a daily basis. Also critical is a user representation, on who's
behalf the network is built.

If you would like to participate, it would help us greatly to obtain
from you a short position statement in advance. ou will have an
opportunity to discuss these views during the workshop.  Please also
let us know whether you are interested in giving a short presentation
of your views to the workshop attendees. From the workshop we will
construct a report to be distributed to the participants, the National
Science Foundation, and other interested parties. We will also attempt
to make position statements and pointers to relevant information
available via the Web in advance, as we receive them, to allow for
information distribution prior to the workshop.

The workshop will be held in January 1996 at the San Diego
Supercomputer Center. Attached you will find attendance information.
Please confirm with us by responding to this note whether you expect to

We have submitted a proposal to the National Science Foundation to help
those participants who do not have own funding to cover the travel for
the workshop. If this is the case for you, please contact us for
information about possible availability of such funds.

Hans-Werner Braun, and Kimberly C. Claffy (SDSC),
and Mark W. Garrett (Bellcore)

Workshop co-chairs