RIPE 40

RIPE Meeting: 40
Working Group: IPv6
Status: FINAL
Revision Number: 2

Please mail comments/suggestions on:



R I P E 4 0 P R A G U E
RIPE-40

Minutes - IPv6

DRAFT Minutes of the IPv6 Working Group (1st version)

RIPE 40, Prague, October 2nd, 2001
Chair: David Kessens
Number of participants: 83 participants

A) Status of 6Bone
(David Kessens, presentation available from http://www.kessens.com/~david/presentations/)
* Noted growth in IPv6 sites
* Better understanding of the system
* Growing number of inet numbers
* More addresses are being allocated
* Addition of two countries - 'co' & 'cu' (presumably Colombia and
Cuba)
* Number of queries are up
* No questions were asked

B) Addressing on cable systems
(Pascal Julienne)Reminder: example is a typical site in one city/TV
and phone connected to one server
* Addressing scheme of cable networks are more complicated when
connecting cities i. Routable sites are more complex
* ii. One WAN for one city, one WAN for another, and so on.
* iii. As you grow cable networks in a country, you can tie up
several cities with different servers, making routing scheme more
complex

Why is IPv6 attractive for cable networks?
* i. Unique address requests from large number of end users
* ii. IPv6 Anycast address may provide effective web traffic between
service points

Host address
* i. IPv4 vs Mac based addresses

1. Idea is not to renumber but use existing addresses
* ii. Digital TV

Shows service-based aggregation scheme/aggregate by service
* i. Two possible ways

a. Prefer services over geographical areas
b. Geographical areas over service
In one case, you can manage all services from one department
In the other case, you can manage different cities
c. Merged scheme has yet to be examined thoroughly

IPv6 vs IPv4 interaction
* i. IPv6 or IPv4 or use of a tunnel?

Protocol stack
* i. Multicast aspect in cable network
* ii. Most countries use DVB today
* iii. Modifications need to be made for the future
* iv. SMPTE 305M is IPv6 compatible but not best for Europe

Q: Is there any IPv6 vendor support for cable equipment ?!?
A: - End-user's PCs are ahead
- Set-top boxes and head-end need IPv6 implementation
- Digital TV and telephony support unknown

Where do we go from here?
Internet on TV is not going well in Germany but speaker believes it is
possible

Q: We have a problem with Anycast.
A: Anycast works only in the routers and not in the service
Anycast needs to be investigated to overcome problems

Q: You can connect to the Internet via the private
topology/connectivity of IPv6, cable systems, /48 systems?
A: it depends how it is managed What I show is not an answer of how to
do it but an example

Q: Along the lines of Francis ... you can't give an end-user one IP
address..., one customer in Paris...

A: this is just an example

I am working with Euro and US cable set box manufacturers. In one
month I can give you names of several companies who are developing
equipment
A: can we get the names now? (laughter)

C) Global IPv6 Routing Table
(Gert Doering, presentation available from
http://www.ripe.net/ripe/meetings/archive/ripe-40/)
* This will not be scientific but impressions;
something to show and discuss
* I will give an overview of the topic
* What is the routing table?
* Protecting addresses
i. One big difference to new address: everyone gives transit to
others
ii. Everybody gets connectivity
iii. Underlying topology is different from IPv4
* Some continents have different structures, leading to some
interesting observations

i. (Shows Saturdays' numbers)
ii. 6BONE
iii. Most important for the future is a /35
iv. Shows distribution graph of total number of prefixes since
Saturday
v. Shows graph separating from 6BONE and registry base
vi. Shows new slide of RIRs, AS, and AS paths
vii. Overall there are 171 addresses participating in the table?
viii.Average number of Hops is so similar via IPv6 vs IPv4
ix. Observations of 'crap' in the table
x. Problems of address length and of too many hops
xi. Find out what router is causing them and fix
xii. Lazy aggregation - Cable & Wireless example
xiii.Need to be more careful of what is put into the IPv6 table
otherwise it will look like IPv4 table, full of 'crap'
xiv. Invalid AS numbers (shows example of Mexican ISP)

Comment: I saw AS numbers being stolen (gasps/laughs)
* I want to make IPv6 work, so if people use 'crap', I will use that
For now, I just observe weird things and call them to make them
stop
* Conclusions: it works, people are hooking up to IPv6, it is still
experimental, people are cooperative, interest is growing, even
with the illegalities (stealing, etc.)

Comment: on filtering - we can discuss this at next meeting
* We need to be careful of which packets are used for certain
circumstances
* Shows slide of references

Comment: We have 250 routes; it would be interesting to have a
registry as we did before.
* This is something to be discussed later

Q: You mentioned what could we do to improve? Anything we can do here
(at RIPE?)
A: We can look at the table and if we see irregularities; just ask
them for the reason; many may not respond; but if everyone just talks
to their neighbor.ask if it is done on purpose..

Q: I am tempted to place this on the agenda topic next time. Q: I want
to ask people what addressing scheme they use for
point-to-point links
A: I use 127's. I have never seen any problems using them.

D) Discussion on RPSL and IPv6 routing policies
(Florent Parent, presentation available from
http://www.ripe.net/ripe/meetings/archive/ripe-40/)

Florent discussed whether the implications on RPSL regarding RPSL. He
discussed the various options of adding attributes and classes.

The decision was made to form a maillist to discuss the RPSLng:
rpslng _at_ ripe _dot_ net

E) Pan-European IPv6 IX Backbone
(Jordi Palet, presentation available from
http://www.ripe.net/ripe/meetings/archive/ripe-40/)
* Remarks on receiving funding for IX backbone
* Began project in April 1983
* Project still under negotiation

i. Hopefully, we'll start on December 1st
* Correct pronunciation: 'Euro6IX' as in 'EuroSix'
* Build a native IP structure

i. Want it to be 'open'
* Will receive funding from the EU
* "Chicken and Egg" problem overcome
* Gaining real IPv6 experience in the real world

i. Simulates real commercial networks
* Euro6IX Goals and Objectives

i. Consortium numbers

1. Main Telco's represented by their research arms
2. Universities
3. Sponsors are international ("and it's just starting")

ii. Network hierarchy
iii.Network map
iv. Infrastructure set-up
v. Work package summary
vi. Research activities
vii.Users/User group involvement
viii.Access Usage policy

1. For EU funding, we must stay clear of conflicts

F) IPv6 in European Region
Input from the audience requested:
IST/6NET IPv6 testbed
(Bernard Tuy, presentation available from
http://www.ripe.net/ripe/meetings/archive/ripe-40)

Q: What is the difference between Euro6IX and IST?
A: Membership is different
Q: But the goals seem to be the same.
A: Yes, most are the same. Both projects will incorporate a lot and be
common but will never be equal.
Money has been put into IPv6 development and that's important.