APCCIRN-016

1992.12.21

>To: ccirn@Csa1.LBL.Gov

Cc: iepg@aarnet.edu.au, Michau@urec.fr (Tel 44 27 42 59),

Frode.Greisen%uni-c.dk%cise.cise.nsf.gov@Csa1.LBL.Gov,

rcollet@icm1.icp.net (Robert D. Collet)

Subject: circuit upgrade/re-homing to Europe, FYI

Reply-To: goldstein@nsf.gov

Date: Fri, 20 Nov 92 10:09:50 EST

From: sgoldste@cise.cise.nsf.gov (Steve Goldstein--Ph +1-202-357-9717)

Status: OR

Dear Colleagues,

As we approach several rounds of regional CCIRN meetings, and as topology

issues appear on several agendas, I wish to offer some circuit updating

information. To those of you close to the planning for individual links,

this may not be news, but I want to make sure that all colleagues are

informed.

Best wishes as various of our cultures approach holiday seasons.

Steve G.

----------------------EUROPE---------------------

NSF participates in three major circuits between North America and Europe via

th e International Connections Manager (ICM) [for NSFNET] Program: London,

Stockhol m and France [Sophia Antipolis -> Paris transition]. All three are in

the proce ss of being upgraded in [base] bandwidth to T1. However, each is

different in i mportant aspects of organization from the others: Stockholm: now

768 kbps, with commercial service providers in Sweden and the US paying for 256

kbps of the sha red capacity (256 kbps access circuit in Stockholm limits

commercial utilization ). Scheduled for upgrade (February 1993 or

earlier--order has been placed) to T 1, with commercial providers remaining at

256 kbps. NSF and NORDUnet are fundin g entities for non- commercial capacity.

London: now 768 kbps, channelized with hard-multiplexing (MUXing) to two

channels: "Networking Research" at 256 kbps and "Infrastructure" at 512 kbps,

both nominal values and modifiable with changes in MUX settings.

Infrastructure channel operated by NASA and JNT for NASA mission and NSF/JANET

general infrastructure traffic; Networking Research channel operated by

DARPA and UCL/DRA for a variety of uses. Infrastructure channel extremely

congested. Upgrade to T1 expected before end of 1992 (maybe as early as

end of November). Will be re-channelized into Networking Research (256 kbps),

NASA/Mission (256 kbps) and Infrastructure 1 Mbps, again all nominal values.

Infrastructure channel will be operated by JNT and ICM. Link is funded by

several agencies in US and in UK, teaching us all new things about cooperation

and coordination each time we face the challenge of making changes.

Paris: 512 kbps now installed, and at this writing, Renater (France's

new Research and Academic network) is migrating French networks to the

connection; 128 kbps link to INRIA/Sophia Antipolis will be abandoned

when cutover is completed (weeks). Link will be upgraded to T1 by end

of February, 1993. NSF and Renater ar the funding entities. Several

French ministries participate in Renater funding.

All three links (just the Infrastructure channel on the London link) are

coordinated with Ebone, and are thought of by NSF as links to Europe via

Ebone in one way or another, though cooperatively funded by national

entities in Europe at this time. This in no way diminishes NSF's appreciation

of the contributions of partners in Nordic countries, UK or France. But,

rather, it is a statement of NSF's strategy of utilizing its limited resources

for investment in consolidated links to Europe and counting on Europeans to

route the traffic within Europe. In 1993, the members of the Ebone

Consortium of Contributing Organisations (ECCO) may take over some or all of

the funding of the European half-circuits for these three links. But, at

this time that is only speculation.

All three links have been coordinated extensively at the engineering level

among all participating parties, as well as at the policy level among

funding entities. The overlapping national network/Ebone Operations

and Action Teams/IEPG memberships have contributed mightily to the

coordination at the operational levels.

---------------- LATIN AMERICA------------------

Latin American and Caribbean countries continue to meet in a variety

of groupings and settings to cope with the challenges of regional

cooperation. It is still premature to fold this into the CCIRN picture,

as there is little overall consensus, amid growing indications of islands

of local consensus, which should be nurtured.

NSF has asked ICM to install a router in Homestead, Florida at the teleport

of PanAmSat (satellite company used by many Central and South American

countries to reach the US and Europe) to permit easy and well-engineered

connections to the global infrastructure. the router would have been

operational by now except for the aftermath of Hurricane Andrew. We have

been informed by the local carrier, Southern Bell, that it may not be until

7 January 1993 that they make the final cross-connection in the local loop

that will connect PanAmSat to Sprint's (ICM) long lines. ICM will take

the traffic directly to it's ICM POP in the Washington, DC, area for

connection to the Global Internet. Costa Rica and Ecuador are just waiting

for the cross-connect to make their connections (meanwhile, Ecuador

connects temporarily through University of Miami and SURAnet). Peru is

close to connecting this way, too, and Colombia, Bolivia and Uruguay have

expressed interest in connecting this way. Costa Rica is working with

neighboring countries on the Central American isthmus (e.g., Nicaragua,

Panama, Guatemala) to connect them to the Costa Rican hub to form a regional

network. The organization of American States has worked with Caribbean

nations to form a UUCP-based CUnet as a precursor to full connectivity.

Reports from a Latin American Net Workshop in Merida (just ending now)

are that there is an encouraging spirit of regional cooperation among

Latin American networkers. So, we anticipate further progress.

----------------------------PACIFIC-----------------

NASA and PACCOM have been prime actors here, and NSF has little to add.

Malaysia will be connecting to ICM this month. Indonesia is in the

running to receive a World Bank loan which would include Internet

connectivity, but realistically, we do not expect them to connect for

about a year. Connection will not necessarily be to ICM, and this item

is for info only.

There are reports of interests of additional agencies in Japan to connect

to the Global Internet, and one, reputedly all the way to Europe (physically

through the North American continent, but independent of US infrastructure

and thus helping to extricate participants from the policy issues that

surround transit traffic). While NSF has been consulted about some of the

issues, presumably as a courtesy, NSF is not a participant.

--------------------OTHER-----------------------------

Turkey, Kuwait, Saudi Arabia and United Arab Emirates are in

various stages of connecting via ICM. ICM has been instructed

to advise all inquiring parties in regions close to Europe to

give careful consideration to connecting via European nodes,

but some countries insist on connecting directly to the United

States (at their own expense), and that is their privilege.

Egypt seems to have listened to this advice and appears to be

pursuing a single connection through Paris rather than individual

connections to the US and to France. The T1 from Ebone/Renater/Paris

might add incentive for such considerations. NSF hopes so, and we

acknowledge the fine cooperation of Renater ane Ebone in these matters.

South Africa is re-homing their connection from a 9.6 kbps connection

to an Alternet customer to a 56 kbps connection to the ICM/Washington

router. The circuit has been in acceptance testing for over a week, and

technical problems continue to be isolated and resolved with the cooperation

of Sprint and Telkom (ZA).

NSF continues to work with other US Federal agencies to obtain consensus

approval to route IP traffic from the former Soviet Union in the US

Federal nets. It should be understood that Russian IP traffic is already

being routed via the CIX association members. Baltic States' traffic

has been routed in NSFNET since this past summer.

-----------INTEGRATION OF COMMERCIAL SERVICE PROVIDERS---------

NSF has made a conscious and deliberate choice to cooperate with

private ("commercial") service providers in implementing global

connectivity. This takes many forms: the Cooperative Agreement

between NSF and the ICM (Sprint) calls for service rather than

circuits and leaves the door open for link sharing, as long as

NSF is reasonably assured of receiving the service (bandwidth)

it pays for. The capacity sharing on the Nordic 768/T1 is an

example of this. In addition, Malaysia and some Latin American

countries have expressed requirements to co-route research and

Academic traffic and commercial traffic on their links. In that

case, the ICM routers route non-AUP ("commercial") traffic to the

CIX member networks and AUP traffic to the FIXes. The countries

involved are paying for the circuits themselves, though NSF pays

the ICM a "port fee" to manage the connections. Since Ebone is

AUP-free, the ICM will route non-AUP traffic from Ebone similarly.

NSF is open to sharing arrangements on the link to Paris, and

Renater is not opposed to this. This would open the Paris-NSF

link up to other-than-incidental non-AUP use by securing payment

for non-AUP use, as in the case of the Nordic circuit. As regards

the London circuit, there were strong objections among US partners

to partnering with commercial providers on the Infrastructure

channel, though the UK partners were willing to entertain the prospect.

It is possible that Eboen 93 decisions might facilitate mixed use on

these links.

But, that brings us to a fundamental issue: the Internet is witnessing

increased participation by commercial service providers. We delude ourselves

to continue to think that we can do realistic coordination in isolation of

the non-research_and_education networking community. Sven Tafvelin has

made constructive proposals that point in the direction of operational

analogs of IETF and IAB under the aegis of the Internet Society (ISOC).

There may be other approaches worth considering, and it may be practical to

pursue several avenues simultaneously. CCIRN members should note that the

IEPG seems to have recognized this by inviting the participation of

representatives of commercial service providers from all CCIRN areas at

its meeting in Washington this week. Might we follow their lead?