Late minutes from 3rd EOF, Lisboa.
- Date: Sat, 19 Nov 1994 18:31:51 +0100
What? Me? Late? Naaaa ... just a month or two.
My profuse apologies for keeping this to myself for so long. I keep
telling you I'm a lousy minuteman, but you won't listen ...
Find below the first draft version of the minutes from the 3rd EOF
meeting in Lisboa. As usual I'll be happy to receive comments (that
includes spelling and English language) to the list or in private, and
I expect plenty this time, because there is lots of places where I've
had to patch my comments with the contents of my memory, and it's
known to be extraordinary poor! First person to file a comment wins a
beer (eq.)! :-)
I especially ask Keith, Jean-Michel, Erik-Jan, Bernard T, and Peter L
to check the technical descriptions of your respective IX. My notes
are even more foggy there (the mind forms the note, you know ...) than
in other places. E.g., Erik-Jan, I only have the acronym IAE for the
IX in Amsterdam, so I'm guessing wildly at what it stands for.
Please everyone, also check the spelling of your name, I had some
difficulty reading some of the handwriting on the sign-up sheet.
With a second round of excuses, here goes!
Cheers,
/Liman
#-------------------------------------------------------------------------
# Lars-Johan Liman ! Internet: liman@localhost
# Ebone/NORDUnet/SUNET Operations Centre ! BITNET : LIMAN@localhost
# Royal Institute of Technology, Sweden ! Voice : Int +46 8 - 790 65 60
#-------------------------------------------------------------------------
#----------------------------------------------------------------------
# DRAFT - DRAFT - DRAFT - DRAFT - DRAFT - DRAFT - DRAFT - DRAFT - DRAFT
#----------------------------------------------------------------------
3rd EUROPEAN OPERATORS FORUM
MINUTES
Date: September 12, 1994
Place: University of Lisboa, Lisboa
Chair: Peter Lothberg roll@localhost
Minutes by: Lars-Johan Liman liman@localhost
Participants:
Erik-Jan Bos (EJB) SURFnet Erik-Jan.Bos@localhost
Stephan Biesbroeck (SB) BELNET stephan@localhost
Rob Blokzijl (RB) Nikhef k13@localhost
Christophe Chaillot (CC) France-Telecom chaillot@localhost
Yves Devillers (YD) EUnet Yves.Devillers@localhost
(representing Glenn Kowack EUnet)
Sabine Dolderer (SD) DE-NIC dolderer@localhost
Havard Eidnes (HE) NORDUnet Havard.Eidnes@localhost
Gilles Farrache (GF) IN2P3 farrache@localhost
Elise Gerich (EG) Merit epg@localhost
Jan Grvanric (JG) CESnet thjg@localhost
Willi Huber (WH) SWITCH huber@localhost
Jean-Michel Jouanigot (JMJ) CERN jimi@localhost
Daniel Karrenberg (DK) RIPE NCC Daniel.Karrenberg@localhost
Andreas Knocke (AK) DE-NIC knocke@localhost
Petr Kral (PT) CESnet psl@localhost
Ingrid Ledererova (IL) CESnet il@localhost
Lars-Johan Liman (LJL) Ebone liman@localhost
Peter Lothberg (PL) Stupi roll@localhost
Keith Mitchell (KM) PIPEX keith@localhost
Arnold Nipper (AN) Xlink nipper@localhost
Alain Pegorari (AP) France-Telecom pegorari@localhost
Marc Pichon (MP) Transpac pichon@localhost
Jean-Michel Planche (JMP) OLEANE jmp@localhost
Franck Pradal (FP) France-Telecom pradal@localhost
Juergen Rauschenbach (JR) DFN jrau@localhost
Andreas Schachtner (AS) EUnet/DE `whois schachtner`
Willem van der Scheun (WvdS) SARA scheun@localhost
Milan Sterba (MS) RIPE Milan.Sterba@localhost
Nigel Titley (NT) BT Nigel.Titley@localhost
Bernard Tuy (BT) RENATER tuy@localhost
Peter Villemoes (PV) NORDUnet Peter.Villemoes@localhost
Wilfried Woeber (WW) ACOnet woeber@localhost
1. WELCOME
Peter Lothberg opened the meeting and wished everyone welcome.
2. MINUTEMAN AND AGENDA
Lars-Johan Liman was volunteered to take minutes.
The agenda sent out by PL was approved with the following change:
Point 5 "Election of EOF Chairman" was moved to follow point 9.
3. EXCUSES
Excuses were forwarded from Fergal Suipeil, HEAnet, Frode Greisen,
Ebone, Glenn Kowack and Per Bilse, EUnet, Eric Malmstrom, Transpac
Sweden, ...
4. PREVIOUS MINUTES
The minutes from the previous meeting were approved, including the
changes suggested on the mailing list.
5. ELECTION OF EOF CHAIRMAN
(This point was moved to 9.5.)
6. TRAFFIC AND ROUTING EXCHANGE AGREEMENTS BETWEEN NSPS
PL brought forward the need for a common language when it comes to
discussions on IXs. The terms used have to be well defined, so
that all parties mean the same thing when they are negotiating IXs.
The need for a "Serive Provider's Handbook" was recognised. The idea
is to be able to inform new players on the market how the Internet
exchange systems work and what is expected from an SP, what the
technical terms are, etc.
It was felt that it would be a good idea to have something like a
"generic IX agreement form" to be used as a template for partners
wanting to formalise an agreement. KM, PL, and PV volunteered to
draft such an agreement form, using PIPEX's standard agreement and
a PSI standard agreement as input. This draft should be circulated
to the list one month before the next meeting.
7. EXPERIENCE OF EXISTING EXCHANGE POINTS IN EUROPE
LINX - The London Internet Neutral eXchange (Keith Mitchell).
*) D-GIX type of architecture.
*) All major SPs in the UK are connected.
*) LINX handles only UK internal traffic.
*) LINX is located at Telehouse in Docklands. Telehouse have a lot
of telecom capabilities, but do not act as an IP SP in the UK,
which makes them neutral and suitable for the task.
*) Rules:
- SP must have its own connection to the world outside UK.
- SP must have IP service within the UK with public pricing.
- SP must connect to the LINX with at least 64 kbps.
- SP provides its own router and pays all the way for its line
to LINX.
- SP must have a funding of at least GBP 2,000.
Dante - Ebone traffic exchange, CERN - Amsterdam (Jean-Michel Jouanigot).
*) Not a normal IX, since there are only two partners. There is no
contract and no "local club".
*) Statistics are provided per application, per network, per host.
IAE - Internet Amsterdam Exchange (Erik-Jan Bos).
*) Operational since 1 year.
*) Consists of an ethernet at the WCW.
*) Started as an interconnect between SARA - CWI - Nikhef.
*) The ethernet is soon to be upgraded to an FDDI ring.
*) Anyone is welcome to connect. Just bring your router and your
line.
*) "No rules" for connecting, but for exchanging traffic you need
bilateral agreements with the organisations you wish to exchange
traffic with.
*) For 1994 SURFnet provides the facilities (space, power and
cooling, etc).
*) There are discussions going on for 1995. Probably there will be
a small fee for connecting, to cover housing costs.
The Paris D-GIX (Bernard Tuy).
*) No formal system yet.
*) Will be housed by France Telecom.
*) Discussions on the requirements are going on, but will probably
be something in line with
-) There will be a housing fee.
-) SP has to be BGP4 capable.
-) SP pays all the way for its line to the IX.
-) SP has to promise not route other SPs' traffic.
*) If no agreement can be reached with France Telecom on how run
the facilities, it will be moved elsewhere.
The Stockholm GIX (Peter Lothberg).
*) Consists of two Catalyst ethernet/FDDI switches.
*) KTHNOC provides rack space for one box per SP, up to and
including the size of a Cisco 7000, as well as UPS power,
cooling, and a connector to the preferred media (ethernet or
FDDI).
*) All Swedish telecom SPs are connected, as well as several
foreign.
*) The Swedish government has provided money to run the facility
for 5 years.
*) Rules are:
-) SP provides router and pays all the way for its line to the
IX.
-) SP router is managed exclusively by the SP, so that each SP
can set its own policy.
-) You have to peer with at least two other SPs at the IX.
-) There is a route server that emulates routers, implementing
the policies registered the routing registry.
Statistics were discussed at length. Two points of view were found
to be of high importance:
*) There should be privacy on an IX. No one should have to tell
anybody else what kind of traffic they exchange with other SPs
at an IX.
*) It might be tricky to engineer the IX itself properly, if you
don't have access to accurate statistics, showing where there
might be bottle-necks etc.
It was concluded that that different IXs will have different
policies. It will be market driven.
8. ROUTING SYSTEM, CIDR
PL pointed out a CIDR problem.
Customer Foo, multihomed,
has all 199.0.0.0/8.
...............................
____________ ___ ! . ___ .
! NSP ! ! ! ! . ! ! .
_____ ! E3 !-----! R !----!-.---! X !---- C1 = 199.0.1.0 .
! !----!____________! !___! ! . !___! .
! IX ! ____________ ___ ! . ___ .
!_____!----! JBnet ! ! ! ! . ! ! .
! 9.6 kbps !-----! R !----!-.---! Y !---- C2 = 199.0.2.0 .
!____________! !___! ! . !___! .
! . .
...............................
If Foo announces 199.0.0.0/8 to NSP all packets go to *either* to X
*or* to Y. You have to send specific routes which get propagated,
even though they don't make sense in a global scope (e.g., at the
IX). This situation cannot be reflected in the routing registry for
the moment.
Another dangerous situation. *If* NSP aggregates the routes for
Foo, but JBnet does not, all packets heading for a network that has
more specific routes will go through JBnet instead of through NSP.
CIX routing restrictions.
According to KM the CIX wants to implement a policy so that if you
want your packets to be routed over the CIX you have to be a CIX
member. They want to use a routing registry to achieve this. The
question was asked if anyone knew how this was intended to be done,
what systems were to be used, etc. The answer was not known.
RIPE route server removal.
DK announced that the route server at RIPE will cease to be later
this autumn. The PRIDE project is going away, and there will be no
one to run it. According to PL, the MAE-East++ will have at least
one. It is also possible that ISI/Merit will set one up.
DK recommended people to make sure they were not relying on the
service.
JMJ suggested that people working with this should get in touch
with the RIPE Routing Working Group.
PL pointed out the need for a common format for all route
servers/routing registries, so that data could be exchanged between
them.
CIDR progression.
Status: depressive!
Some SPs are using non-CIDRised routes to implement local
policies. SPs should *always* announce CIDR routes to the outer
world. The cost for implementing local policy should always be
carried by the one implementing it, i.e., the SP!
9. EOF CHARTER
It was decided to send out the EOF charter to the mailing list
again, and to finalise it at the next meeting.
9.5 ELECTION OF EOF CHAIRMAN
It was decided to ask for nominations on the list. KM forwarded a
message from Peter Dawe (PIPEX) that he will happy to stand as
nominee for the chair.
The election of a new EOF chair was postponed to the next meeting.
10. PRESENTATION OF MAE-EAST++
PL briefly presented how the new MAE-East will be designed.
11. AOB
There was none.
12. NEXT MEETING
It was decided to hold the next meeting in the London/Cambrige area
on November 25, 1994. PIPEX/Keith Mitchell kindly volunteered to
host it.
#----------------------------------------------------------------------
# DRAFT - DRAFT - DRAFT - DRAFT - DRAFT - DRAFT - DRAFT - DRAFT - DRAFT
#----------------------------------------------------------------------
|
you are here: RIPE -> RIPE Maillists > Archives
