Attendees:
Kevin Barron, Institute for Theoretical Physics
Steve Francis, Communication Services, CalREN-2 Technical Advisory Council
Nathan Freitas, Electronic Art and Technology Lab
Jim Frew, Alexandria Project
Ben Humphrey, Computer Science
Elise Meyer, Physics
Larry Murdock, Psychology & Biology
Kevin Schmidt, Campus Network Programmer
Vince Sefcik, Communication Services, CalREN-2 Business Advisory Council
Bob Sugar, Physics
Robert Nideffer, Art Studio & Alexandria
Kevin Lovette, Art Studio & Alexandria
Unable to Attend:
Stephen Pope, Center for Research in Electronic Art Technology
CalREN-2 Implementation Group
Bob Sugar explained that the CalREN-2 Implementation Group (C2IG) is a
subcommittee of the Campus Network Committee (CNC). C2IG is comprised of
members of the CNC Backbone Engineering Group (BEG), Communication Services,
and representatives of the four research groups included in the CalREN-2
proposal, plus two additional research groups. C2IG will advise the
CNC regarding the CalREN-2 network, and the CNC will then advise the
Executive Vice Chancellor.
The intercampus portion of the CalREN-2 network is planned to be operational
on January 1, 1998, and the goal for C2IG is to have a working campus network
at that time to carry that traffic. It is expected to take two months to
order, install, and test the required networking equipment, so we should plan
on having a final design and cost recommendation at the CNC for approval prior
to November 1, 1997.
The six research projects will be the primary consumers of initial bandwidth.
But the long-range goal is that UCNET will move to CalREN-2. Decisions will
need to be made regarding the division of authority between the research groups
and communication services. There is some initial funding (120K) for backbone
equipment, but more may be required, so part of our recommendation may be to
ask for funds from EVC Crawford.
Elise will put up a web page for the group at
http://www.physics.ucsb.edu/calren2/. Vince requested that access should be
limited to ucsb.edu only. The official mailing list for C2IG will be
calren2-tech@ucsb.edu. There is also a business and administrative oriented
mailing list that is called calren2-admin@ucsb.edu. You will automatically
be added to the tech mailing list. If you want to be added to the admin
mailing list, send your subscription request to calren2-admin-request@ucsb.edu.
Overview of the CENIC Organization and the CalREN-2 project
CENIC: Consortium for Education Network Initiatives in California (soon to
be Corporation)
The members of CENIC are UC, CSU, USC, Stanford, and Cal Tech. Stuart
Lynn from UCOP helped spearhead this organization. CENIC submitted the
CalREN-2 Proposal for an NSF Connections grant. Projects like CALREN-2 will
be mostly autonomous and self-supporting Cal State is not yet part of CALREN-2.
CENIC will have a board comprised of 3 UC, 3 CSU, and 3 private
education/research institutions. The UC boardmembers will rotate through
the campuses.
CalREN-2: California Research and Education Network - 2
An NSF Connections 96 grant winner in the High Performance Connections for
Research and Education Institutions and Facilities category. It is a two-
year grant that runs from July 1, 1997 to June 30, 1999. Phase I (which is
what has been funded by the NSF) is to build a northern gigapop containing
the Northern UCs and Stanford, and a southern gigapop containing the
southern UCs (except UCSD), USC, and Cal Tech. These gigapops will be
interconnected by the vBNS. CalREN-2 will provide the official UC connection
to the vBNS.
CALREN-2 is consistent with, but not automatically a part of, Internet2.
Campuses must join I2 individually; currently only UCOP, UCB, UCD, and UCLA
have joined. The annual cost of membership (membership fee, backbone network
usage and campus network build-out) is currently $25K, but it will soon be
$500K. CALREN-2 and Internet2 will both use the vBNS as the top-level
backbone.
There are several advisory councils for CalREN-2. Vince represents UCSB on
the Business Advisory Council. Steve Francis represents UCSB on the Technical
Advisory Council. There is no UCSB representation on the Academic Advisory
Council. Any interested individual can join either the AAC or the TAC, whose
mailing list is cenic-tpg@ucdavis.edu. [Update: the AAC is being chaired by
Gayle Byock gbyock@conet.ucla.edu.]
CalREN-2 Network Topology
Note:
OC-48 = 2.4 Gigibits per second
OC-12 = 622 Megabits per second
OC-3 = 155 Megabits per second
- Phase I:
A Southern Gigapop with OC-48 connections between UCLA, USC, ISI, Cal Tech,
UCI, and UCR. UCSB will be connected to UCLA at OC-12. A Northern Gigapop
with OC-48 connections between UCOP, UCB, UCSF, and Stanford. UCD will
be connected to ? at OC-12. The rings will be connected via the vBNS.
The southern ring will be connected to vBNS from UCLA at OC-3, and the
northern ring will be connected to the vBNS from UCD at OC-3. [Update:
vBNS connections may be OC-12 instead of OC-3.] UCSD & SDSC is not part
of Phase I, but they will be accessible via UCNET at 10Mb/s or via their
connection to the vBNS at OC-3.
- Phase II:
Connect the gigapops with OC-12.
- Phase III:
Connect UCSD & SDSC to the gigapops at OC-12.
Since CALREN-2 uses Sonet, phases II and III will incur an inter-lata per-packet
tariff which is estimated at $6.5M/yr.
UCSB's current offcampus network connections cost $150K/year and we have a
4Mb/s SMDS connection to the UCNET. Resnet has a 1.54 Mb/s T1 connection
to MCI.
Research Group Application Bandwidth and Service Guarantee Requirements
A brief description was given for five of the six research projects.
- Elise:
High Energy Physics is transferring huge datasets produced either by
detectors at Cornell or SLAC, or produced by simulations. They are
currently transferring 20GB/week via tape; they would like to transfer
20GB/Hour via the network. They are also flying faculty and graduate
students to these sites. They would like to reduce their travel costs
by using video-conferencing instead.
- Ben:
Klaus Schauser is doing parallel computing. The SCI (Javelin?) project
is currently saturating a 30Mb/s link. Kevin Almeroth will be doing
multimedia and MBone multicast video.
- Jim:
ICESS has 20 PIs doing remote sensing. They do a lot with the SDSC.
ES&M is doing tele-collaboration, distance learning, and developing a
prototype for remote sensing service organizations. (esip, uci connection?)
Alexandria Digital Library will be a web service organization for the
entire UC. They are also going to be mirrored by SDSC. They are also
going to be doing auto failover. (cdl?)
One issue for the above projects is where they will connect to the
new network, since they have facilities located all over campus.
- Kevin Barron:
ITP will be doing "radio free physics," i.e., multicast itp seminars.
They want to put up the Public Lecture Series voice and view graphs.
They expect that they will have an audience all around the world with both
low- and high-bandwidth capabilities. They do have a quality of service
requirement for the real audio.
Larry noted that he was digitizing Bio lectures.
- Nathan:
The Electronic Art Technology Lab will be doing online journals, and
Multi-user 3D spaces for their "online public spaces" proposal. They
will need a connection to SDSC, and they think that their traffic will
be bursty.
The EAT Lab currently doesn't have fiber running to their building (Art).
Vince mentioned that he had been asked to come up with a cost estimate for
doing that.
Campus Backbone Design Issues
We will need to determine what sort of equipment is needed and where it should be
located. This will depend on the research group locations, existing equipment,
and network requirements. The research groups will need guidance on how
best to exploit their limited number of connections, and help characterizing
their bandwidth needs.
CalREN-2 Intercampus Issues
One of the things that will need to be determined is whether any of our
research group projects have requirements that have an impact on intercampus
network design. For example: How to implement end-station to end-station
Quality of Service requirements? Do we need to be concerned with NHRP
(Next Hop Router Protocol)?
Action Items for Next Meeting: July 25, 1997
For the next meeting, each research group is to identify:
- current equip
- current applications
- estimated bandwidth needs
Back to Main CalREN-2 Page
EMM
