Attendees:
Ben Humphrey,
Roger Lenard,
Elise Meyer,
Larry Murdock,
Kevin Schmidt, and
Jason Simpson
Unable to Attend:
Kevin Barron,
Nathan Freitas,
Steve Francis,
Jim Frew,
Stephen Pope,
Rich Prohaska,
Vince Sefcik, and
Ed Mehlschau,
Responses to Our RFI and Query
We have received 5 responses to our RFI. None of the vendors that have
responded currently have equipment that meet the critical ATM standards
required to do ATM Quality of Service.
We received 5 responses to the query that we sent to the cenic-tpg, ucatm-tech, and Susan Estrada.
None of the respondents were installing an ATM backbone on their campus.
Most were using big bandwidth to satisfy their users' needs.
Some were looking into RSVP for providing some sort of QoS.
I have posted all of these responses to our web page under the "Questions and
Answers" section.
We briefly discussed Layer 2 QoS strategies like "ATM QoS" versus Layer 3
strategies like RSVP. RSVP is not even a standard yet, and its capabilities
are less than those "theoretically" available with ATM QoS. Our conclusion
is that neither Quality of Service mechanism is currently available in a
standards-based solution. We think that RSVP will gain greater usage because
it is aimed at IP traffic.
Fiber Issues
We then discussed the state of campus fiber. We know that we have a problem
with Arts, because there is no fiber to that building. We also have a problem
because most of the research group locations only have access to multimode
fiber. Only the Library has access to singlemode fiber. The BEG is currently
drafting a recommendation destined for the CNC about completing the deployment
of fiber to all buildings on campus. We thought that the CalREN-2 project
might provide further motivation to the campus to put in more fiber and to
also go back and pull single mode fiber to the existing rings that don't
currently have it.
CalREN-2 Backbone Topology
We haven't specified the location of the backbone equipment yet, because we
didn't know what the equipment would be or how many interfaces the equipment
would have. According to the fiber allocation information made available by
Steve Francis, and distributed by Elise at the September 25th meeting, there
doesn't appear to be a problem with fiber going to the CalREN-2 research
groups.
New Campus Backbone Design
Our original backbone design had the following features:
- Backbone ATM switches connected at OC-12 (622 Mb/s).
- Research Group connections at OC-3 (155 Mb/s) with the
option of paying more to get an OC-12 connection.
Since ATM equipment doesn't currently provide the Quality of Service
capabilities that we are looking for, we are now looking to our new backbone
to just provide big bandwidth. ATM will probably not be the most cost effective
way to do this, so we are considering other technologies to interconnect our
backbone equipment and to provide high speed connections to our research
groups.
We developed "Contingency Plan 1" for discussion. This
topology has the following features:
- There would be some sort of FSB (Fancy Spiffy Box) located at
Public Safety (to connect to CENIC switch), North Hall (to connect to Ellison,
Givertz, Library & Music), and Bio 2 (to connect to Broida, ITP & Engr I).
- Connections from the local FSB to the research group would be
some type of switched 100 Mb/sec. We assume that this would be a parallel
network, not a replacement for the existing FDDI backbone.
- Connections between FSBs would either be via multiple low speed
links using MFP (Multiple Fiber Pairs), such as vendor proprietary port trunking, or via inverse multiplexing.
The connections between FSBs could possibly be upgraded to make use of high
speed interfaces, either OC-12 Packet over Sonet (OC-12 is limited to 300
meters on multimode fiber), or Gigabit Ethernet (limited to 260 meters on
multimode fiber).
Possible FSBs would be the Cabletron MMAC+ or the Cisco Catalyst
5500. The FSB should have "Functional Modularity" or "Modular Functionality."
V. Action Items for Next Meeting: October 15, 1997
- Elise will put the RFI and Query responses up on our web site.
- Elise will write up our current direction in order to poll the
research groups to see if they would be satisfied with a
connection of switched 100 Mb/s instead of OC-3 ATM, and if they
needed dedicated 100Mb/s through the campus network.
Back to Main CalREN-2 Page
EMM
