Please mail comments/suggestions on:
- content to the Chair of the working group.
- format to webmaster _at_ ripe _dot_ net.
IPv6 Working Group Meeting RIPE38, Amsterdam Wednesday January 24,
A. Administrative stuff - appointment of scribe - Vesna Manojlovic,
RIPE NCC - attendees list circulated - agenda bashing (David
B. Status of the 6bone (David Kessens)
C. DNS and IPv6 (Roy Arends, Nominum)
D. IPv6 support for the root nameservers (Francis Dupont)
E. IPv6 demonstration in the terminal room (Francis Dupont)
2nd part 16:00-17:30
F. European Union projects regarding IPv6 (Gordon Lennox, European
G. Revised ietf bcp based on ripe/apnic input (Randy Bush)
H. IPv6 allocations for 3G mobile operators / SubTLA IPv6 Allocations
or Who is afraid (Niall Murphy)
I. Are /35's enough as initial allocations ?!? (Bernard Tuy, *****,
FR and Stuart Prevost, BT, UK)
J. Developments/initiatives regarding IPv6 in the RIPE region and
beyond (input from the audience)
K. Future plans for the working group (input from the audience)
Z. AOB --- [david, i know you'll cut out the agenda part above ;) ]
=A= Administrative stuff (David Kessens) - appointment of scribe -
Vesna Manojlovic, RIPE NCC - attendees list circulated - agenda
bashing - many items added to the original agenda - issue of
Thomas Trede forwarding ietf announcements to ipv6-wg. How manu
people want that to continue? More then not - so, we will keep it.
=B= Status of the 6bone (David Kessens)
Summary: steady growth of all activities.
Server moved to another location, due to the small power outage in
=C= DNS and IPv6 (Roy Arends, Nominum)
Topics: New Records; Zone file examples; BINDv9; Issues; Work to be
Issues: A6/DNAME cause overwhelming dns traffic, referrals can cause
lookup loops, part can be signed => renumbering will be more expencive
due to nserver changes costs;
Jim Reid (Nominum): BINDv8 will never do A6 and DNAME records.
Question from David to audience: Do you use BINDv9? Yes, and it's
Frances: We need the resolver. Jim Reid: lightweight resolver is
there, to be used and tested. Lightweight resolver deamon can do it,
the a conventgional stub resolver can not do that!! We need the
input/feedback from the community. Certain vendors probably will not
include it with their distribution.
David: I have BINDv8 adn AAAA records on my server. I'd like to
upgrade my server version... Roy: BINDv9 is stable. Jim: If you wan
to do it for IPv6 - you just have to use BINDv9. Just for upgading,
you will have to change all your scripts . Warning: BINDv9 server is
slower - threaded; maloc; depending of platform - by factor 10; only
if you have 100/second you will experience problems. 15 ms -> 100 ms
to answer query.
Check www.uug.org for half day course by Numinum on upgrading to
Q (moheman): why is it not posible for cashe server to specify the
subset of servers on which it runs? Roy: right now - any or none. the
Q: You said that partially signed records are possible, and that it's
not clear if they are then secure or not. My opinion is that - if in
the proces of resolution you step in the non-signed zone, then you can
not declaire that it is secure resolution. A: examples: rfc-2535,
Jim: A6 (and DNAME) are evel for DNS. please kill this record in
IETF. there is no benefit. potential serious problems, aditional load
on all nservers, people make mistakes with referels, ...
=D= IPv6 support for the root nameservers (Francis Dupont)
We need IPv6 support on root name servers.
There will be another discussion in the DNS working group about this.
[David, help - what was this?? DId I get it right??]
Jim: Resolving in IPv6: plans of nominum in BINDv8 stub resolver will
resolve ipv6. there is no resolver library in BINDv9, but the one
from v8 will be included. there will be suport for native v6
Francis: If you are IPv6 only, you need dual stack ... there are no
serverss for TLD and root servers. There is a need for that. There are
some patches, thou.
The same problem is for DNSSEC, but that's another story.
... only to requester which use ipv6, and who use ipv4 to get A and
lookup goes to ipv6 server, and it returns ....
... aditional records:
ccTLD registrars: feature requests for BINDv9, let us know. it will
just require some programing.
This is continuing in dns-wg.
[Sorry, David, I didn't get all of this!!!]
=E= IPv6 demonstration in the terminal room (Francis Dupont)
3 boxes in the room in front of registration desk 1. tunel 2. ??
3. freebsd bindv9 on one of them, in order to have dns.
=F= European Union projects regarding IPv6 (Gordon Lennox, European
Projects: 6INIT, LONG, AQUILA, SEQUIN, GCAP, NetGate, DriVe (9 mil
EURO), WINE GLASS, MOBY DICK, BRAIN, SUITED,
Q: anybody who is a part of some of those projects, and has to say
something about it?
Jurgen: - sequin - impressive list of projects, and it is good that EC
is suporting them. it would be good to see the detailed results of
the v6 projects. Gordon: we now are trying to make more coordination
between those projects.
2nd part 16:15-17:30
=G= Revised ietf bcp based on ripe/apnic input (Randy Bush)
IAB/IESG recomendation [slides]
[why was this refused in ARIN? what happens next?]
=H= IPv6 allocations for 3G mobile operators / SubTLA IPv6 Allocations
or Who is afraid of RIPE-196 (Niall Murphy, ******, UK)
Doing A, B and C for startups takes one year - before you can get
SubTLA ; bootstrap criteria expires after 100 in total, or 60 in one
region, so startups may not make it to apply to IPv6 under bootstrap
Q (david): how many are today? A: 25, 11, 24
Q (Niall): what entities are involved into writing ripe-196 and
similar documents? he would like to get involved! A: Mirjam: you have
all been involved in creating it - document has been under discussion.
It is living and draft document, obviously we can change things, if
there is concensus.
Q (Paul Wilson): What is the problem with expiring bootstrap phase?
A: Criteria will get even stricter. Q: You would have similar
criteria, just peering with 3 IPv6 peers. A: There are not enough
=I= Are /35's enough as initial allocations ?!? (Bernard Tuy)
Bootstrap phase is coming closer to it's end. Suggestion is to change
policy document. 13 bits are not enough for building hierarchy.
Reverse delegation of IPv6? - nobody from NCC knows if this is
Inupt from Stuart Prevost, BT, UK
Stuart showed several possibilities to buld hierarchy with the
existing allocation size, and few suggestions for the change of
Timothy, RIPE NCC: (TO Bernard) You said that no-one in RIPE NCC knew
about the IPv6 reverse delegation nserver. You haven't asked everybody
- you haven't asked me. B: I've asked 3 perosons. T: There are more
then 50 people working for RIPE NCC. David: That does not scale, if
he has to ask each one of them. Manuel: we will get back to you. B:
Is there some kind of RIPE NCC IPv6 expertise group, which can answer
this question? David: This is a question of misscommunication. Leo:
Is the question - do we have an ipv6 server, running ipv6 stack and
having reverse delegation? That is the question.
(Discussion about the size of the initial allocation)
Paul Wilson: /35 = 13 bits of NLA space equivalent of /19 what is the
fundamental difference? A (David Prat?) not possible to build
hierarchy. in ipv6 we only need to worry about agregation, not about
Paul: after /29 LIR goes directly to /16. David Kessens: /48 proposal
came later; we need to review the /35 proposal.
Chair: This looks like recomendation to Registries to give more space.
(to increase the size of the initial allocation?)
Bernard: We need to re-write the document.
Mirjam: We should concentrate on this particular proposal. We should
work together - RIRs, IETF, operators.. David: We (RIPE) are ahead of
other forums. ARIN is also working on that... Jurgen: The process is
too slow. Recomendation is 6 months old, and no rewriting of policy
document occured. We have many problmes: hierarchy, assigments to
cusomers, small companies not being able to get ipv6 at all.. We can
not wait for ARIN to come up with common view. David: We might
organise a task force, representatives, and actually get some work
done. Mir: As in the case of IPv4 policies, policy changes need to be
proposed by the community. Personally, I would like to see one unique
Chair: Details in the draft are not clear for start up providers,
process needs to be speed up; there is second issue of /35 - most
people in the WG agree with that.
Are there any smaler ISPs that are against that? (No)
Then, the recommendation from the RIPE community is that something
needs to be changed. Who is going to make progress on this? RIR were
waiting on input from ARIN.
Richard Jimerson (ARIN): We have new IPv6 working gropu discussing
this issue. The closure is expected by the April meeting. If the
proposal comes to change the size of initial allocation, two things
can be discussed at the same time.
Somebody: GPP network... 3gpp ipv6 will be running on the private
Bernard: I am going to write something down.
David: How can we speed up this proces? It is clear that there is
Randy: RIRs are doing fairly good job in representing operators. Mir,
Joao .. should do firmer job in presenting this. /35 is really your
Chair: I hope we will be able to discuss this further in the ARIN
=J= Developments/initiatives regarding IPv6 in the RIPE region and
beyond (input from the audience)
J.1. Stuart: www.bt.co.uk/ipv6/ ; www.uk6x.com
J.2. Manuel, RIPE NCC SW department: reminder of the tool releised -
=K= Future plans for the working group (input from the audience)
Randy: Is /48 a good idea? Yes, that was answered last time.
Paul Wilson: Chainese were assigning /48 to the university, but then
they started asking for /48 for the deparment. Problem is the fixed
boundary per site, and the definition of "site".
David: Some wording should exist to describe the intention...
News: IPv6 tutorial to be organised in Bolognia.
There were no AOB.