APCCIRN-038
APCCIRN-038
1993.10.07
CCIRN Meeting August 23-24,1993
Bodega Bay, California, USA
FINAL MINUTES (21.6.94)
[a summary of each issue is contained in the last paragraph under
that heading]
FORMAL ACTIONS TO BE TAKEN FOLLOWING THE MEETING
1. Peter Kirstein agreed to find out who can update us on TINA.
2. Barry Leiner and Simon Holland will write a paper discussing
the subtleties of the meaning of the mission of the CCIRN and to
subsequently review the Terms of Reference and Bylaws to make
sure their research/education focus issues are therein addressed.
3. Greg Chartrand agreed to see if the Russian DOE plan could be
circulated outside DOE and if so to send it to Bill Bostwick for
distribution to the CCIRN mailer.
4. Glenn Ricart agreed to send the CCIRN statement on the IEPG to
the IEPG co-chairs. (Done 8/26/93; see statement at end of
minutes.
5. The two remaining co-chairs will liaise with the IEPG on their
position on multiple GIXes.
6. Tomaz Kalin will work with the RIPE NCC to create a position
paper on funding the IANA as a top-level NIC supported by the
(currently 3) regional NICs. This paper will be coordinated with
the IEPG.
OPENING
Host Co-Chair Bill Bostwick opened the meeting at 9:15 a.m. with
introductions and a review of the agenda.
MINUTES OF THE BRUSSELS MEETING
The minutes were approved unanimously. Co-chair Bostwick thanked
Simon Holland of the Commission for the hospitality.
UPDATES
North American (NA) Updates
Peter Ford from Los Alamos National Laboratory on loan to the
National Science Foundation reported on the status of the NSFnet
solicitation. A substantial number of proposals were received;
panels are now being setup to review the proposals. "It would be
nice to get an award made by the end of the year. This is a
target we've set in our minds." George Strawn has gone back to
Iowa State, but is still part-time. Priscilla Huston of Rice
University is the new NSFnet Program Director. "There is no doubt
we will see significant change in the network in the U.S." It
will be more competitive and providers will be more global. Glenn
Ricart asked about international connectivity and transit traffic
after April 1994. Peter Ford replied that ANS will not be turning
off its network in April, and "the NSF will work very hard to
make sure international connectivity will be maintained." Kees
Neggers noted that everything the NSF has done has been very
open, but that the rest of the world doesn't really understand
the consequences. The engineers are only beginning this
discussion today, but until the awards are made,
it isn't clear with whom negotiations should be conducted. Steve
Goldstein said that "International connections made at the level
of the NSFnet backbone will be able to get to their intended
places. Those made at that level will be taken care of. This
applies specifically to the GIX and the West Coast. For example,
the GIX may be bridged to the West Coast. ... The International
Connections Manager agreement with Sprint has 2.5 years to run."
Kees observed that Europe no longer has a body with which to
negotiate connectivity improvements. He asked if NSF would be
willing to act as that body. Peter Ford replied that NSFnet is
attempting to set up a structure that will work both domestically
and internationally.
Bill Bostwick announced Nico Habermann died of a heart attack two
weeks ago; he was chair of the Federal Networking Council.
The ATM service for DoE and NASA Science Internet is "very very
near resolution" and will bring in the 45 Mb/s part of the
Internet.
Steve Goldstein announced that Peru signed a contract with PamAm
Sat during iNET93. Bolivia is very close. Columbia sent the head
of their NSFnet equivalent to iNET and we expect that Columbia
will be on within the next 3-6 months. Venezuela is already
connected, but they may re-connect via PamAmSat in Homestead.
Argentina is presently disconnected due to their local telephone
company. Another group in Argentina may create a second link.
Several countries in Central America may form a star around Costa
Rica. Honduras may get its own connection to Homestead in
adddition to connecting to Costa Rica. Costa Rica will divide
into three subdomains for government, education, and commercial
(RACSA).
The CANARIE project is now officially named CANARIE. It has
received government funding; the agreements to disburse the
funding should be completed in the next few weeks. There should
be T1 links across Canada within a year.
Glenn Ricart announced that 7 U.S. regional networks have formed
an organization called CoREN--Corporation for Regional and
Enterprise Networking. They will create their own inter-regional
connectivity and will be doing an ATM test network in the next
few months. A production timetable depends upon the results of
the NSFnet solicitation. The regionals forming CoREN are: SURAnet
(southeastern US), NEARNET (northeastern US), NYSERnet (New
York), SESQUINET (Texas), MIDnet (middle US), CICnet (Great Lakes
states), BARRNET (California), and Northwestnet (northwestern
US).
Peter Kirstein asked if the carrier-based experiment in
international ATM interconnection (34 Mb/s) (TINA) was working
with the U.S. internet. No one knew of such cooperations, but the
question will be re-raised when the IEPG is present tomorrow.
European (Euro) Updates
Kees Neggers reported that RARE and RIPE are very stable. The
RIPE NCC is doing very well.
Enzo Valente reported on the HEPnet consortium connecting the
high energy physics laboratories in Europe and Russia and the
United States ESnet. Plans are to upgrade the lines to follow the
needs.
An operational unit has been set up under the name DANTE,
as a Limited liability company in Cambridge, England.
Howard Davies is one of the two joint general managers;
Tomaz Kalin is one of the directors.
EBONE now serves as both an important interconnect backbone
in Europe and also as a transcontinental collector/distributor
in Europe. About half the EBONE traffic is destined for the U.S.
and Asia-Pacific. It will continue for this purpose.
EMPB (European multiprotocol backbone) provides both X.25 and IP
at up to 2 Mb/s. The X.25 service has been in place since October
1992. The IP service passed its pilot service test in June 1993
and is presently in operation. Many networks now plan a transition
from Ebone to EMPB. The EMPB service is a result of the COSINE
project and is based on a contract between PTT Telecom (Dutch PTT)
with RARE (to be assigned to Dante). The contract defines function
and performance and provides guarantees. The EMPB service is operated
by Unisource, a subsidiary of the Dutch, Swedisch and Swiss PTT's.
The term "EuropaNET" is to be the term for the complete network
service Dante will provide including EMPB. Work is underway to see
how EMPB can replace EBONE for the intra-European traffic. If use
of EMPB becomes dominant, a transition of international links from
EBONE to EMPB will be contemplated.
EBONE is planning next year's operations. Dante is a member of the
EBONE group and provides a gateway to EMPB. There are some
countries not planning to use Dante and EMPB. Christian Michau
noted that some of these countries want to extend EBONE at least
until EuropanNET has proved itself; others such as France want to
see EBONE continued for the long term. More planning will be done
at the September meeting of the EBONE group. EBONE is willing to
maintain competition with EMPB. Kees Neggers noted that those
wishing to coordinate with Europe will have to work with both
DANTE and a new consortium of 6-7 countries intending to continue
with EBONE including France (RENATER).
Simon Holland expressed concern with the split and urged a single
organizational point of European contact. Sven Tafvelin noted
that getting the European cost structure changed to encourage a
unified network should be a high priority of the Commission.
Howard Davies expressed Dante's long-term ambition to win the
support of all Europe to represent them internationally.
DANTE is currently leasing a virtual private network from
Unisource. Later, PTTs may provide other virtual private
networking options that DANTE may choose to utilize according to
Peter Kirstein.
Steve Goldstein wished noted for the record that "what NSF has
done [in restructuring US connectivity] has not been done any
more unilaterally than what Europe has done [in considering
changing the infrastructure]." Kees Neggers wished to object
noting that the U.S. process has accepted some architectural
principles that, as the process continues, may limit the possible
outcomes. In Europe, the discussion is just starting, and the
CCIRN and all others are invited to provide their input; no
commitments to a process or conclusion have yet been taken. Kees
Neggers does not question the NSF's intent to improve the U.S.
networking situation; he is simply concerned about the possible
outcomes that may follow from their procedure. Howard Davies
noted that the European changes have not affected US connectivity
at all, while the US changes have substantial consequences for
Europe.
Another important development in Europe is that the COSINE
Project has been concluded after three years. Dante is now setup
to provide services. There is no successor for the policy group.
The Commission with Norway and Switzerland has tabled a new
Eureka project called EuroCAIRN to setup a high capacity
infrastructure including gigabit networking. There is also
potential participation by Hungary, Slovenia, and Russia. The
next meeting on EuroCAIRN is September 7. Firm expressions of
interest have been received from all eligible members except
Ireland. It will promote 34 Mb/s within Europe and extend its
interest beyond connectivity into applications.
On the 24th of May, the Hilog met. This is the High Level
Officials Group intended to set high level policy for European
networking. It is as yet unclear to whom Hilog will permanently
report.
Another new development is that NATO is interested in providing
support for networking in central and eastern Europe. The support
is via RARE; the first visible action will be meetings in
Budapest in October and next summer in Moscow. The person
responsible in NATO was previously responsible for Commission
activity in COSINE.
Simon Holland announced changes in Commission DG13. The whole
ESPRIT project has been moved to the industry director general.
That leaves DG13 focus on telecommunications and networking and
support for EuroCAIRN.
Bill Bostwick asked for continuing updates on TINA. Peter
Kirstein took an action to find the right person.
Asian-Pacific (AP)
Kilnam Chon gave the AP update. Fifty to sixty people
participated in an APCCIRN meeting following iNET. An experiment
to form a counterpart of the Internic and RIPE NCC coordinated in
Asia-Pacific (APNIC) is going to last until next spring. At that
time, a decision on its future should be made. The main function
is assigning domain names and network numbers. They will also
start a routing registry soon.
Kilnam Chon noted the need for discussion between Asia and Europe
on issues such as link coordination between them as most of the
transit traffic over the USA is the transit traffic between Asia
and Europe. The APCCIRN decided to include the commercial service
providers. At the next meeting in December, the 8-10 commercial
companies in AP will be asked to make presentations.
Another important issue in AP is local language support.
Curtis Hardyck discussed the Pacific Neighborhood Consortium. It
consists of universities, national libraries, and some education
ministries. It was started by the presidents of 17 institutions
meeting in Seoul two years ago. Since then, it has grown to 40
institutions in 16 countries. The goal is to develop a database
of machine-readable databases. The next meeting is January 17-18
in Hong Kong and will discuss how the databases can be made
available on the Internet if there are no restrictions on use, or
at least available to PNC members if there are restrictions on
use. The PNC is member-supported and contributes the resources of
its library and computing people. Hardyck can provide a brochure
to those interested.
Guam, Macau and Vietnam are being connected to the Internet. The
Phillipines and Indonesia are getting connections via the World
Bank and UN Development Program. The AP region has 80% of the
developing countries, but only 10% of the attendees at the iNET
developing countries workshop were from AP.
APCCIRN covers the countries from the Pacific to India and
Pakistan now. Countries in central Asia and the Middle East are
not covered by either APCCIRN nor EuroCCIRN at this moment. We
need to cover these countries as many of them are not networked.
IAB Updates
Barry Leiner provided an update on iNET93 which was attended by
860 people from 91 different countries. The iNET Workshop for
Developing Countries had 126 attendees from 66 countries. iNET94
wil be in Prague (inet-jenc-request@rare.nl) June 13-17, 1994
jointly with JENC5. ISOC is attempting to acquire a class "A"
relationship with the International Standards Organization (ISO).
NSF has made a small grant to VITA to videotape sessions at the
iNET workshop. If something comes out of this, training videos
would be made available to the community at large. Non-exclusive
rights were given to ISOC.
ISOC Updates
Tomaz Kalin, who is a board member of ISOC, reported on its
status. ISOC will begin supporting the IETF this year with
$300,000. iNET95 may be in Singapore. Only one of the six elected
board members came from outside the United States; this is an
issue of concern. Regional elections are being discussed. An
Executive Director is being sought. A contract for the
Secretariat is to be negotiated with CNRI.
There are still some discussions about whether the IESG is part
of ISOC. ISOC is discussing liability issues it may acquire.
International Cooperation Board (ICB) Updates
The last meeting of the ICB discussed privacy-enhanced mail and
workstation conferencing among the members, but did not raise new
policy issues.
COMMERCIAL / GENERAL SERVICE PROVIDERS (evolved into: ROLE OF THE
CCIRN)
[To skip the discussion and go to the summary conclusion, look
for the last paragraph in this heading.]
In Brussels, the discussion focused on the formation of the CCIRN
by the research and education communities and the need to retain
this focus. Barry Leiner remembered the purpose of the CCIRN
decided in Brussels was to serve the research and education
community. The core membership in the CCIRN was to be those
people whose job was to provide network resources to the research
community. Clearly, the U.S. National Science Foundation and RARE
are among them.
Peter Kirstein observed that India and the Arab countries have
chosen to be connected via commercial carriers and have therefore
opted out of CCIRN. [Comments from Kilnam Chon indicate that this
statement needs to be verified as it is contrary to what the
APCCIRN believes to be true.]
Simon Holland believes there is a need for a research/education
community networking coordination body worldwide whether or not
it relates to the Internet.
Kees Neggers agreed that the CCIRN should not be limited to a
single protocol stack.
At the ISOC board meeting, there was discussion about whether the
ISOC should form an International Operations Board (IOB) or
sponsor a meeting of the likely players. Neither of these
happened.
Barry Leiner suggested the primary role of the CCIRN is to
provide a communications vehicle for fostering international
coordination. In his view it should not be a policy body per se.
Simon Holland suggested that if international policies are not in
place, it will be very confusing for all players. Bill Bostwick
noted the CCIRN has been successful in helping international
groups share policies with one another.
Greg Chartrand sees a large coordination role for science users
of commercial networks.
Barry Leiner provided a three part justification for the
uniqueness of research and education networking. (1) Needs often
outstrip commercial availability (2) Free networking is often
needed to stimulate interdisciplinary and intercontinental
cooperation (3) Networking researchers need testbeds.
Glenn Ricart noted that coordination of ISO level 2 links is no
longer needed within the research and education community alone
because of the growth of the commercial networks. The IEPG may
continue to work in this area (Greg Chartrand). The CCIRN could
evolve either toward inclusiveness in level 2 routing and involve
commercial vendors, or "move up the protocol stack" and work on
research/education issues closer to the application layer. In
Brussels, we seemed to choose the latter approach.
Steve Goldstein recommended that Barry Leiner and Simon Holland
review the CCIRN Terms of Reference in terms of their excellent
comments. They agreed.
Kilnam Chon and Barry Leiner had a discussion agreeing that
networks that serve education and research will also carry
commercial traffic, but that the CCIRN focus is on the research
and education users. Bill Bostwick reminded us we don't really
represent the users, but only their support structure.
James Hutton suggested we should consider passing all lower level
issues down to IEPG and all higher level issues up to an
international version of Hilog. CCIRN tries to progress the
issues of the research community without referring them to
cumbersome government resolution mechanisms.
[Summary of the discussion:]
Barry Leiner stated the summary for the group. "The CCIRN is
about coordinating the provision of networking services to the
education and research community. The CCIRN has formed the IEPG
for the purpose of coordinating the engineering aspects of that
coordination mission. The CCIRN recognizes that coordinating
global engineering requires the participation of the broader
community providing services to the research community including
commercial network providers. The CCIRN needs to be concerned
with management and policy issues revolving around the provision
of service to the education and research community." Simon
Holland and Barry Leiner took an action to write a small paper
discussing the subtlety of these points.
FUNDING THE GLOBAL OPERATIONS OF THE NIC
[Monday discussion]
Many members of the CCIRN are concerned that the root of the
number assignment space is presently concentrated in a single
country's government (the United States).
Steve Goldstein promised that no action will be taken without
full consultation of the international community, and that
delegate registries should be employed.
Barry Leiner suggested the best solution to this issue may be to
ignore it because the root has little or no power if there is
strong delegation.
Rob Blokzijl emphasized the need for well-defined international
policies. He believes the best solution is for a true global NIC.
Tomaz Kalin agreed that there must be an international root, even
if its cost is very small.
Kees Neggers believes that the growth of IP in Europe will be
greatly retarded if the IP version 4 address space is not put
under the control of an international group/party. Barry Leiner
believes that however desirable this is, there is no experience
in operating any global root service.
Bill Bostwick stated his personal opinion that the U.S.
Government does not want to give up control over the IP
networking space.
Rob Blockzijl said that it is painful to see how the APNIC and
RIPE NCC NIC are avoiding the issuance of class "B" network
numbers in accord with international agreement, while the U.S.
regional NIC (which also happens to be the top-level NIC) doesn't
follow these rules and is issuing class "B" numbers without
commensurate restraint. Without a separate root, there is no one
to whom to take these concerns.
Kees Neggers asks all CCIRN members to go home and ask their
government to internationalize the control of IP so that we have
an open IP networking environment. This includes number
assignment and the IETF. If any government disagrees, they should
bring back that disagreement and it will be discussed.
Kilnam Chon agrees and suggests a position paper be drawn up to
this end.
Steve Goldstein and Bill Bostwick agreed that the issue is larger
than the CCIRN; it involves the general provider community.
[Tuesday discussion]
Glenn Ricart opined that the policy issues which still need
resolution are globalizing the root of number space and setting
policy and organization of the GIXes. Kees Neggers reported that
IEPG recommended handling the root issue by globalizing the IANA
(Internet Assigned Number Authority); he is pleased with that
direction.
Tomaz Kalin stated that "we want a symmetrical hierarchical
structure with a top level number authority (perhaps IANA), and a
series of regional NICs (currently 3) which will finance the
IANA." He was asked by co-chair Bostwick and agreed to produce a
proposal for discussion purposes and coordinate it with the IEPG.
JOINT MEETING WITH IEPG
[Steve Wolff of the NSF joined the CCIRN meeting.]
REQUIREMENTS
(Guy Almes excused himself.)
Greg Chartrand explained the requirements for DOE and NASA to
China and Russia. The three regions of interest in Russia are the
Moscow area, St. Petersburg, and Novasibirsk. The top priority is
Novasibirsk. DOE is putting in a link to St. Petersburgh to
address the fusion and high energy requirements.
The Russian plan is through the DOE and is being studied by the
Dept. of Commerce. The China plan should be finished by the DOE
this month.
There is a DOE High Energy Physics link between SLAC (Stanford
Linear Accelerator) and IHEP/Beijing. ESnet is not routed to the
link at present, but the China plan will address this.
Co-chair Bostwick asked if the Russian plan could be distributed
outside DOE. Greg agreed to take an action to see if it were
available and to send it to Bill Bostwick.
In response to a question about routing for the Russian links,
Greg replied that DOE has been told by Commerce that they can
route traffic delivered by a link someone else has installed.
Rob Blokzjil explained that there are three commercial providers
in Russia. All institutions, including non-profits like the
Kurchatov Instititute, use these commercial services.
Steve Goldstein explained that NSF's goal was infrastructure
growth. The (privately financed) International Science Foundation
(ISF) has been active. There is a direct 64 Kb/s link since first
of June from Sprint Networks in Moscow and the ICMnet router
connected to the GIX. The Sprint Networks operation is joint
between Sprint and the Moscow telegraph company. There is a Cisco
router in Moscow with 12 synchronous ports and a terminal server.
GLASnet is already connected to the terminal server. SOVAM will
come up soon; they also have their own link to San Francisco.
Other sources in Moscow have ordered dedicated (19.2 kbps analog)
links to the router. The NSFnet backbone has been unable so far
to pass Russia and China, but the GIX and CIX announce this
traffic to the commercial networks. ISF also plans to assist with
connections from Novosibirsk and with infrastructure in
Novosibirsk, St. Petersburg and a national backbone for Russia.
Also, with Internet demonstration and development projects in
Ukraine, Belarus and other states formerly in the Soviet sphere.
In addition, the ISF has agreed to fund a satellite link at 64
Kb/s from the Russian Academy of Sciences to the High Energy
Physics Institute in Helsinki. It will be available for general
infrastructure. The rates are favorable because the satellite is
turned off at night to save batteries.
ISF and RELARN have agreed to help fund the lion's share of a
Russian-built fiber optic backbone in the phone company conduits
from Kurchatov Institute of Atomic Energy, through Moscow and the
Russian Academy of Science Presidium Building, through the Sprint
Building, IKI (Institute of Space Research) which will have the
NASA and DOE link, and the M9 switch. The installation will be
done by ROTEC(?) by the end of October and may be in operation as
early as the first part of next year with 15 institutions.
In Novosibirsk, a group had generated 80% of the financing for a
link to Sweden. ISF will provide the other 20% and it will be
installed by October.
The St. Petersburg Ioffe Institute has a proposal into ISF for
local loop funding to match up to the DOE link to Moscow.
ISF is helping Russia look at a national backbone.
In Ukraine, a small program is looking at PC based routers. The
ISF sponsored 21 people to the developing countries workshop.
ISF/NSF has been contacted by a group from Belarus to connect to
Minsk.
The Turks who have a link to the GIX will extend service to
Azerbaidzhan.
There is also a proposal from Georgia, but they don't have power
all day.
Estonia has three or four links out that ought to be
consolidated; Mats Brunel is taking the lead to help there. DFN
is helping with a link into Latvia. ISF will help Lithuania out
with workstations to be used as network servers. In Latvia, Riga
has problems with local wiring and may need radio modems to gain
connectivity.
Milo Medin reported NASA is seeking a general export license for
Internet data. NASA is trying to pull a 256 Kb/s leased line into
IKI and then route it without any restrictions with the export
license.
(Guy Almes returned.)
IEPG REPORT
IEPG Co-Chair Geoff Huston gave the IEPG report. As reported at
the last IEPG, it is no longer feasible to coordinate individual
links. The emphasis is on creating structures that will scale the
routing problem. The route server experiments at MAE-East are
being watched closely.
The registry function is an integral component of policy
integrity and the major support vehicle for external routing.
IEPG suggests wider dissemination of the role of the registry
function within the provider environment.
Peter Ford raised the problem that some commercial providers
shield the identity of clients; this makes it difficult to solve
end-to-end problems or seek out the perpetrators of security
attacks.
Geoff Huston said we're already at the stage where network
providers must either decide to run CIDR or point default outside
their network. Providers must have registry information and CIDR
tools.
Peter Lothberg said that today's routers have a limit of about
60,000 paths. A path is a way to get to a network. Steve
Goldstein reported that 15,000 networks are now routed. Peter
Ford said the growth under CIDR should stay under the growth rate
for memory and router capability. BGP4 needs additional memory,
and turning on BGP4 requires additional memory in many routers.
Guy Almes reported that ANS expects to do BGP4 and CIDR with
aggregation by October on the ANSnet T3 that supports the NSFnet
backbone.
Geoff Huston reported that the MBONE has grown to more than 500
sites. Milo Medin told the group that Sun is coming out with a 30
frame per second MPEG board that has chewed up the LANs at Sun.
Everyone should be aware that additional bandwidth, routing
tools, and limited resources will result from widespread use of
video.
Geoff continued that registry support is required for CLNP
coordination. There are NSAP issues including the breakup of
country NSAP prefixes. Static tables will have to be replaced
with dynamic routing tables.
The CIDR issues are (a) existing registry roles are not operating
consistently with CIDR guidelines. It will be expensive to
deaggregate CIDR routes if allocation is not done carefully.
Effective use of CIDR will require reclamation and renumbering of
the existing old growth forest of numbers.
The IEPG does not intend to devise and enforce rules for service
providers, but does intend to continue work on a document
describing the roles that should be fulfilled by service
providers. Barry Leiner suggested the document would, like the
host requirements document and the router requirement document,
list the functions that should be provided by the service
provider. Daniel Karrenberg said he has started a document
describing current practices that should be extended.
Geoff Huston reported that the IEPG is already functioning as an
internet operations activity in the context of the entire
internet. It should operate in its own right, not under the
CCIRN. There is much that would have to be set up quickly, but
there is a consensus position to move toward a role in the
Internet as a whole. The CCIRN can still send work requests to
IEPG that will be treated seriously.
Barry Leiner gave the CCIRN position. The CCIRN created the IEPG
because it was needed, and the CCIRN is pleased with the results.
Furthermore, the CCIRN recognizes that the provision of
networking service to the research community comes from service
providers who have broader objectives. In some sense, this is why
the IEPG was created as a separate organization and encouraged to
invite commercial activities. Barry further suggested the ISOC
might either be a home or help find a home for the IEPG.
Steve Goldstein wished the IEPG well and suggested they create
their own parent organization. Milo Medin suggested the IEPG
should be an ad hoc organization without a parent; alternatively,
if left under the CCIRN, it wouldn't be bad because the CCIRN has
largely left the IEPG alone.
Steve Wolff noted that the IEPG has always listened to voices
beyond the CCIRN, so what's new? James Hutton added that the
CCIRN will always be a parent in the sense it created the IEPG,
and that IEPG is welcome to live in a room in the CCIRN's house
as long as it is convenient to do so. But everyone recognizes the
IEPG is an independent group that can talk to those whom it
wishes. Hans-Werner Braun said the problem is that there is a
perception that the CCIRN provided controls and limits to the
IEPG; the divorce aids with the perception issue.
Bill Bostwick said the issue is not perception, but reality. If
IEPG will continue, it needs funding and support. The CCIRN is
willing to support it in this way even while the IEPG has far
wider contacts and responsibilities than a connection with the
CCIRN would suggest.
Barry Leiner suggested getting a charter from the CCIRN but
operating globally. Peter Lothberg said there are others who
should participate in the IEPG but can't because their parents
don't like the religion of IEPG's CCIRN parent. Guy Almes
appreciates the collegiality between the CCIRN and IEPG. At the
policy level, he urges the CCIRN to take on important policy
issues.
Geoff concluded by asking for support for time for productive
navel gazing about where the IEPG fits in.
[The discussion continued with the CCIRN alone plus Steve Wolff
and Milo Medin.]
Considerable discussion ensued that if IEPG wants more than
voluntary compliance, it will have to have a strong parent. Milo
Medin noted that this issue is taking time and attention away
from important issues and that the current arrangement isn't
broken; neither is the 'breakaway' feeling universal among IEPG
members.
Steve Wolff said the power of ideas is what empowers the IEPG.
When policy issues arise, they will take those issues to their
managers. There needs to be a forum then of their managers to
work out these policy issues. The CCIRN should be a forum for
their managers.
Barry Leiner noted that while the CCIRN may not be the suitable
home for the IEPG, the IEPG should weigh heavily the value of
having a home organization.
Sven Tafvelin is of the opinion that ISOC is the only
international organization suitable for operational issues.
The group agreed on the following summary statement: "The IEPG
was formed under the auspices of the CCIRN to provide global
engineering coordination. The CCIRN wishes the IEPG to continue
to operate independently to make the global internet work. The
CCIRN wishes to continue to liaise with them on policy issues."
CCIRN MEETING
CCIRN ORGANIZATION AND ISSUES TO BE PROGRESSED
Bill Bostwick announced this would be his last meeting, and that
he'd make a recommendation to the Federal Networking Council
about the future of the CCIRN at its September meeting. Bill will
be moving on to other endeavors.
Barry Leiner raised the issue of whether a single chair would be
better than the present three regional co-chairs. Kees Neggers
noted that there is effectively a single chair based on the
location of the upcoming meeting. James Hutton suggested the
co-chairs act as an executive committee between meetings.
There was some consternation expressed about the progress made by
the CCIRN between meetings.
Kees Neggers made the point that there is a great deal of
operational coordination that happens informally because the
CCIRN meetings are held. He noted that detracters don't have
operational roles and don't understand the value of the meeting
to operational people.
GIX
[Monday discussion]
Kilnam Chon asked for a second GIX on the west coast of the
United States.
Glenn Ricart explained two kinds of GIX interconnects: (a) those
where all networks connect to all GIXes (like the FIX, CIX, and
NSFnet NAPs), and (b) those where the GIXes are connected to each
other. In the second case, there is a political problem of
deciding what organization provides the inter-GIX connectivity,
its bandwidth, its funding, and its operation. These issues are
avoided in the first case.
Rob Blokzjil explained two roles of the GIX: (1) exchanging
packets and (2) exchanging routing information. Therefore, some
networks may choose to have a 9.6 Kb/s line to the GIX just to
exchange routing information.
[Tuesday discussion]
James Hutton proposed the CCIRN set a requirement or mandate that
there be multiple distributed GIXes. Bill Bostwick repeated that
the CCIRN has asked the IEPG for a multiple GIX solution.
However, the group is of divided opinion as to whether a report
of single GIX is acceptable or not. The two remaining co-chairs
will liaise with the IEPG on their position on multiple GIXes.
There was a consensus that the CCIRN requests a global
architecture that supports more than one GIX. The co-chairs will
liaise with IEPG.
Geoff Huston was asked to join the meeting to discuss GIX. A
summary of his report follows and it was accepted by the CCIRN.
The GIX solves a routing problem by allowing both bilateral
exchange of routes and also communication with a route server.
The architecture can be extended to have several such
environments. The relative locations and policies of network
identifiers are traded across that structure. It doesn't
necessarily pass traffic. The preference is that the GIX is the
first resort of addressing and the last resort of traffic flow.
To extend to multiple locations, what is going on is
synchronizing the exchange of routing information. To solve this,
one can use a level 2 bridge across two such points and simply
trade BGP over the connections. This is viable if traffic is not
passed. Engineering solutions are being proposed to synchronize
address and policy knowledge across a number of sites.
There is a traffic handling problem that remains. The west coast
structures have a diversity of adopted structures. Transit in a
post-NSFnet world is problematical, and IEPG is looking toward
how a well-architected west coast structure could work in May
1994. Geoff responded that one solution is to purchase a portion
of the Washington MFS Ethernet onto the west coast at level 2.
Therefore, additional GIX points can be created for enough
dollars. Milo Medin asked for clarification, and Bill Bostwick
led a consensus statement that "the CCIRN will deal with transit
traffic as a policy issue as it arises and will also deal with it
if the IEPG tells the CCIRN there is an issue."
THANK YOUs
Kees Neggers led a round of applause for Bill Bostwick's kicking
off the CCIRN. Barry Leiner is also thanked for the preparation
of the logistics. He apologized for the previous night being so
cold and windy.
MEETING FREQUENCY AND NEXT MEETING
It was agreed that the baseline meeting frequency will be once
per year with additional meetings called as determined by the
co-chairs as needed by the workload. Adjacency to iNET is viewed
as desirable by the majority of members. A minority expressed an
interest to have the meeting adjacent with other networking
meetings such as IETF.
The next meeting is tentatively set for June 20-21, 1994 with
location to be set by the European chair.
CCIRN ATTENDEES
"Shoichiro ASANO (NACSIS)" <Asano@nacsis.ac.jp
"Rob Blokzijl (RARE)" <K13@nikhef.nl
"Bill Bostwick (NACCIRN)" <bos@lanl.gov
"Greg Chartrand (SSCL/DOE/ESSC)" <Greg@sscvx1.ssc.gov
"MunHou CHEW (NSTB/Singapore)" <MunHou@solomon.technet.sg
"Kilnam Chon (ANS/Korea)" <Chon@cosmos.kaist.ac.kr
"Jim Conklin (CREN/BITNET)" <Conklin@cren.net
"Howard Davies (RARE/Dante)" <H.E.Davies@exeter.ac.uk
"Steve Goldstein (NSF)" <Goldstein@nsf.gov
"Shigeki GOTO (JPNIC/APCCIRN)" <GOTO@ntt-20.ntt.jp
"Alan Greenberg (CA*net)" <Alan@vm1.McGill.ca
"Curtis Hardyck (Pacific Neighborhood Consortium/APCCIRN)"
<Hardyck@violet.berkeley.edu
"Simon Holland (European Commission)" <SHo@dg13.cec.be
"James Hutton (RARE/JNT)" <J.Hutton@jnt.ac.uk
"Tomaz Kalin (RARE)" <Kalin@rare.nl
"Peter Kirstein (ICB)" <Kirstein@cs.ucl.ac.uk
"Barry Leiner (IAB)" <Leiner@nsipo.nasa.gov
"Takayasu MATSUZAKI (APCCIRN/NISTEP/STA)" <Matsuzaki@vnet.ibm.com
"Milo Medin (NASA)" <Medi@nasa.gov [Tuesday only]
"Sandy Merola (Guest from ESnet)" <Merola@lbl.gov
"Christian Michau (RARE/Renater)" <Michau@urec.fr
"Kees Neggers (RARE/SURFnet)" <neggers@surfnet.nl
"Hualin Qian (APCCIRN)" <QianHL%bepc2@scs.slac.stanford.edu
"Glenn Ricart (FARNET)" <Glenn_Ricart@umail.umd.edu
"Sven Tafvelin (RARE)" <Tafvelin@ce.chalmers.se
"Enzo Valente (RARE/GARR/INFN)" <Valente@infn.it
IEPG ATTENDEES
"Akiko AIZAWA (NACSIS)" <Akiko@nacsis.ac.jp
"Guy Almes (ANS)" <Almes@ans.net
"Hans-Werner Braun (SDSC)" <HWB@sdsc.edu
"Joe Burrescia (ESnet)" <Burrescia@es.net
"Che-Hoo Cheng (UPCC/HARNET Chinese University of Hong Kong)"
<CheHooCheng@cuhk.hk
"Milton Choo (Singapore Technet)" <Milton@solomon.technet.sg
"Bob Coggeshall (Hong Kong Supernet)" <Coggs@hk.super.net
"Richard Colella (NIST)" <Colella@nist.gov
"Peter Dawe (PIPEX)" <Peter@pipex.net
"Tim Dixon (RARE)" <dixon@rare.nl
"Peter S. Ford (Los Alamos +1-505-665-0058)"
<Peter@goshawk.lanl.gov
"Kin Fung (UPCC/HARNET Chinese University of Hong Kong)"
<MingFung@cuhk.hk
"Elise Gerich (Merit)" <EPG@merit.edu
"John Haulker (Tuia)" <J.Houlker@waikato.ac.az
"Farooq Hussaini (SPRINT)" <Farooq@icm1.icp.net
"Geoff Huston (AARNET)" <GIH@aarnet.edu.au
"Daniel Karrenberg (RIPE NCC)" <Daniel@ripe.net
"Sun-Young Ko (KT)" </dev/null
"Peter Lothberg (EBONE)" <roll@stupi.se
"Bill Manning (Sesquinet)" <BManning@rice.edu
"Jun Matsukata (ISAS/SINET)" <JM@eng.isas.ac.jp
"Keith Mitchell (PIPEX)" <Keith@pipex.net
"Jun Murai (WIDE)" <Jun@wide.ad.jp
"Masaya NAKAYAMA (Univ. of Tokyo/JPNIC)
"<nakayama@nc.u-tokyo.ac.jp
"Joo Y. Song (Hananet)" <JYSong@ring.kotel.co.kr
"Bernard Stockman (EBONE)" <boss@ebone.net
"Claudio Topolcic (FEPG/CNRI)" <Topolcic@cnri.reston.va.us
"Jack Waters (SURAnet)" <Waters@sura.net
"Weiping Zhao (NACSIS)" <Zhao@nacsis.ac.jp
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
CCIRN Statement on IEPG Relationship
As the recording secretary for the CCIRN at the Bodega Bay
meeting, I was asked to communicate to the IEPG co-chairs the
following statement by the CCIRN regarding relationships with the
IEPG. .... Glenn Ricart
"The IEPG was formed under the auspices of the CCIRN to provide
global engineering coordination. The CCIRN wishes the IEPG to
continue to operate independently to make the global internet
work. The CCIRN wishes to continue to liaison with them on policy
issues."