Internet Info 1994 March/Internet Info CD-ROM (Walnut Creek) (March 1994).iso/inet/ietf/tewg/tewg-minutes-90dec.txt

Text File | 1993-02-17 | 6.7 KB | 166 lines

CURRENT_MEETING_REPORT_



Reported by Phill Gross/CNRI

TEWG Minutes

Past IETF meetings have typically featured status reports from various
operational networks.  These have included NSFnet, the US DOE Energy
Science netwowrk (ESnet), the US NASA Science Internet (NSI), DCA
Milnet, and more recently, occasional regional networks and European
networks.  These reports have typically been featured during the
technical plenary session.

Starting at the Boulder IETF meeting, the network status reports have
been moved into the Topology Engineering Working Group sessions.  This
is in response to suggestions from IETF participants to make time for
additional technical presentations during the plenary sessions.  This
will also have the effect of providing more time for those interested in
network operations to interact in more detail.

This represents a shift in emphasis for the Topology Engineering Working
Group.  I'd like to consider this an experiment for the next few
meetings, after which, I will poll regular TEWG attendees, and others
interested in network operations, for comments regarding the new format,
and whether there are other operational topics that TEWG (or, perhaps,
another Working Group in the Operational Requirements Area) should
consider.

Dale Johnson (MERIT) submitted the text below to accompany his slide
presentation.

NSFNET Presentation (Dale Johnson /MERIT)


   o NSFNET T1 Backbone 1990:
     An additional node was added to the NSFNET T1 backbone in October
     of 1990 in Atlanta, Georgia, bringing the total number of nodes on
     the backbone to 14.
     The Atlanta NSS is located at Georgia Institute of Technology and
     is connected to the NSSs in Houston and Pittsburgh.  In November,
     the Atlanta NSS passed 164 million packets, which was more than
     that passed by two other NSSs.
   o NSFNET T3 Backbone 1990:
     The real story of the NSFNET is occurring with the current
     engineering of the T3 network.  Merit has a goal of passing some

                                   1






  production traffic on the new T3 network within calendar year 1990.
  T3 NSSs will be located in the eight locations shown, which include
  two new sites in Argonne, Il.  and Cambridge, MA.
  The T3 network is being engineered and built as a totally separate
  peer backbone to the T1 network.  It will have its own AS number,
  and will interoperate with the T1 network using an exterior gateway
  protocol.
o NSFNET T1/T3 Backbones 1990:
  As can be seen in the combined T1 and T3 map (see slides), several
  sites will have both a T1 NSS and a T3 E-NSS (exterior NSS - see
  below).  Packets will be routed between the two backbones at these
  locations.
o NSFNET T1 Architecture:
  This is a diagram of the T1 NSS and circuit architecture as it
  relates to the MCI backbone junction point, or Point of Presence,
  (POP). In the diagram, everything inside the circle is physically
  located at the MCI POP.
  The circles labeled DXC represent MCI backbone Digital
  Cross-Connect switches.  There are two (or more) clear-channel T1
  circuits which run over local loops from the DXC to the NSS located
  at the university or supercomputer site.
  This architecture does allow for redundant circuits from the MCI
  backbone to the site, however an outage at the site not only
  disrupts traffic to that site, but also disrupts backbone traffic
  running through that site but destined for other locations.
o NSFNET T1 Architecture:
  The diagram of the T3 architecture indicates that there will be two
  new types of NSSs at each node on the backbone.  Again, everything
  inside the circle is physically located at the MCI POP.
  As indicated in the diagram, the first new type of NSS, the
  Core-NSS (C-NSS) will be collocated at the MCI POP. The C-NSSs will
  form a backbone infrastructure which will be independent of
  activity at the end sites.  Therefore, an outage at an end site
  will affect traffic only to that site, not backbone traffic
  destined for other sites.
  The second type of NSS, the ExteriorNSS (E-NSS) will be located on
  site at the organization hosting the node, as is done in the T1
  backbone.  The E-NSS will be connected to the C-NSS by a single T3
  pipe.
o Number of Networks Graph:
  The number of networks configured on the NSFNET backbone reached
  2063 by the end of October 1990.  This included 527 foreign
  networks.  The number of configured nets by the end of November
  1990 totalled 2125.  (Chart not available at the time of this
  presentation.)
o NSFNET Monthly Traffic in Packets:
  Traffic in packets for the month of October 1990 was 5.25 billion.

                                2






     This represents a 269
   o Major NSFNET Applications By Packets:
     As is the norm, networked mail applications and file exchange
     accounted for the highest usage of the backbone in September of
     1990.  Interactive applications, again as is the norm, accounted
     for the third highest usage of the backbone.
   o NSFNET - The Reliable Network:
     The NSFNET backbone maintained an average uptime status of
     99.88through September 1990.  Included in this calculation are
     Class One outages only, which means a node site was completely
     unreachable due to an NSS being down.  Planned outages, such as for
     UPS maintenance, are included in this figure.



Attendees

Vikas Aggarwal           [e-mail]
Guy Almes                [e-mail]
Ronald Broersma          [e-mail]
Robert Cooney            [e-mail]
Tom Easterday            [e-mail]
Fred Engel               [e-mail]
Peter Ford               [e-mail]
Vince Fuller             [e-mail]
Phillip Gross            [e-mail]
Tony Hain                [e-mail]
Susan Hares              [e-mail]
Eugene Hastings          [e-mail]
Kathleen Huber           [e-mail]
Dale Johnson             [e-mail]
Philip Karn              [e-mail]
Holly Knight             [e-mail]
William Kutz             [e-mail]
E. Paul Love             [e-mail]
Matt Mathis              [e-mail]
Milo Medin               [e-mail]
Donald Morris            [e-mail]
Chris Myers              [e-mail]
David O'Leary            [e-mail]
Joel Replogle            [e-mail]
Robert Reschly           [e-mail]
Timothy Salo             [e-mail]
Tom Sandoski             [e-mail]
Bernhard Stockman        [e-mail]
Roxanne Streeter         [e-mail]
Kannan Varadhan          [e-mail]
Carol Ward               [e-mail]
Dan Wintringham          [e-mail]
Robert Woodburn          [e-mail]


                                   3