Munich IETF ngtrans-tools & -6bone WG minutes
Bob Fink
RLFink@lbl.gov
Wed, 3 Sep 1997 16:46:47 -0700
6bone folk,
Enclosed are the minutes of the Munich IETF ngtrans-tools and ngtrans-6bone
WG meetings.
Thanks,
Bob
_____________cut
here___________________________________________________________
Ngtrans-tools WG meeting
August 12, 1997
Munich IETF
Tony Hain Microsoft chair (co-chair of ngtrans)
Reported by ALain Durand and Tony Hain
Discussion: ngtrans@sunroof.eng.sun.com
Subscribe: majordomo@sunroof.eng.sun.com
Archive: ftp://playground.sun.com/pub/ngtrans
Chairs: Bob Fink rlfink@lbl.gov
Robert Gilligan gilligan@freegate.net
Tony Hain tonyhain@microsoft.com
Notes for the minutes courtesy of Alain Durand durand@imag.fr
Tools session: Various reports and plans were presented.
Brian Carpenter (IBM, UK) reported that the "6 over 4" draft will be
refreshed with no changes.
Tony Hain (Microsoft, US) noted that the charter wording would be
cleaned up to address concerns about the appearance we are mandating a
scenario, and sent back to the list.
Jim Bound (Digital, US) presented the NNAT concept as a means of
dynamically assigning v4 addresses as needed for primarily v6 nodes.
Goal is to allow primarily v6 work groups to be deployed soon, without
requiring a permanent address in the v4 Internet. Attaining that
requires core applications be available over the v6 stack sooner. Also
requires a tie between the DNS and some form of DHCP service with a pool
of v4 addresses. Some applications may not need or want connections with
the v4 Internet, and should not be required to implement the extra code.
Erik Nordmark (Sun, US) noted that the time to start thinking about
translators to allow v4 only to v6 only has arrived. Some systems may
arrive in the market as v6 only. Erik presented an overview of his ID
(draft-ietf-ngtrans-header-trans-00.txt) and the rules for both
directions. Tunneling v4 in v6 may be required for v4 compatible node
within a v6 only routing system. Another option is a set of stateless
translators that would convert the headers on the fly, but would not
attempt v4 options.
Pedro Marques (Cisco, US) discussed ideas about transparent translation
where hosts were unaware of the translator. Does not require tie to DNS
& DHCP, and Cisco has a prototype running. He will submit an ID to the
list for comment.
Dan Harrington (Lucent, US) reported that the update to the ID
(draft-harrington-ngtrans-dhcp-option-00.txt) would reflect the assigned
number from IANA.
_____________cut
here___________________________________________________________
ngtrans-6bone WG meeting
August 12, 1997
Munich IETF
Bob Fink ESnet/LBNL chair (co-chair of ngtrans)
Reported by ALain Durand and Bob Fink
Discussion: 6bone@isi.edu (6BONE Mailer)
Subscribe: majordomo@isi.edu "subscribe 6bone"
Archive: http://www.ipv6.nas.nasa.gov/6bone/
Chairs: Bob Fink rlfink@lbl.gov
Robert Gilligan gilligan@freegate.net
Tony Hain tonyhain@microsoft.com
--------------------------------------------------------
Agenda
--------------------------------------------------------
1. Introduction and agenda / Fink
2. Status of action items from Memphis / Fink
3. New 6bone registry, oveview and issues / Kessens
4. Backbone planning / Fink, Durand, Davies
5. Test plan for aggregation-based addressing / Fink
6. Operational issues on the 6bone / various
7. Implementations in use on the 6bone / Fink
--------------------------------------------------------
1. Introduction and agenda / Fink
--------------------------------------------------------
Bob Fink convened the meeting, giving a status report on the 6bone.
There are now over 150 sites in 28 countries participating in the 6bone.
BGP4+ has been successfully deployed in the 6bone backbon, with 3
interoperable implementations (Cisco, Digital and Telebit), and is rapidly
replacing RIPng, IDRPv6 and static routes in the backbone.
The 6bone now has its own domain, 6bone.net, through which the 6bone web
pages and registry database is available.
The primary DNS for 6bone.net is located at LBNL, and the secondary name
servers are at RIPE an APNIC.
The 6bone web pages are now available at:
http://www.6bone.net
The mail list is available by sending email to
majordomo@isi.edu
with
subscribe 6bone
in the body of the message.
The agenda was accepted without comment.
--------------------------------------------------------
2. Status of action items from Memphis / Fink
--------------------------------------------------------
Bob Fink reviewed the status of 6bone action items from the Memphis IETF.
2.1 CAIRN Backbone Proposal - Allison Mankin
Allison Mankin will readdress this issue at the December meeting.
2.2 RFC1987 changes to use virtual IPv6 provide ID - Hsin Fang
Hsin Fang removed this item from further consideration as the new Test
Aggregation-based addressing plan supercedes it.
2.3 Aggregation-based Addressing structure for the 6bone - Bob Fink
Bob Fink will present his plan, as issued previously to the 6bone mail
list, under agenda item 5. below.
2.4 I-D "Representing IPv6 Tunnels in RPSL" - David Meyer
This item is now closed as it has been implemented in the new 6bone registry.
2.5 New 6bone Registry - David Kessens
David Kessens will present the status of the registry under agenda item 3.
below.
2.6 DNS for localized 6bone routing registry information - Bill Manning
Bill Manning would like to continue pursuing this idea. He earlier told
the chair that he would come forward with a plan in the near future as to
how he would like to proceed. The chair will close this item and await
further submission from Bill before reactivating it.
2.7 Volunteer for I-D on requirements for new 6bone infrastructure - Bob Fink
Bob Fink reported that he has previously had several offers to help on such
a draft. He noted that when the address conversion and backbone
restructuring of the 6bone is done, he will reopen this action as
appropriate.
2.8 Survey of host and router implementations on the 6bone - Bob FInk
Bob Fink will report on this survey under agenda item 7 below.
--------------------------------------------------------
3. New 6bone registry, oveview and issues / Kessens
--------------------------------------------------------
David Kessens reported that the conversion away from the old ftp-style
6bone registry to the new RIPE-style 6bone registry was completed in early
June. He autmatically mapped over all entries. The status of the new
database is that there are 172 site objects and 130 persons registered.
The query level is at ~200 per day. There are two mirror sites:
whois.nic.fr and whois.ra.net,
David noted that his draft on the registry specification,
draft-ietf-ngtrans-6bone-registry-01.txt
has been updated and it would be processed as an Informatinal RFC.
David also noted that his registry was ready and capable of becoming an
address registry for Aggregation-based Unicast addresses, if it was
desirable.
--------------------------------------------------------
4. Backbone planning / Fink, Durand, Davies
--------------------------------------------------------
Bob Fink spoke about the need to restructure the backbone to minimize
peering, as opposed to full mesh peering, and that the renumbering for
aggregation-based addressing might be a good time to accomplish this. He
also noted the importance of moving to BGP4+ peering.
Alain Durand, of the G6 project in France, spoke on plans to restructure
the G6 collaborators for the coming readdressing required for the move to
Aggregation-based addressing.
Guy Davies, of UUNET/UK, spoke on plans to restructure the UK academic
6bone backbone, both to cleanup the topology and ready for the readdressing
required for the move to Aggregation-based addressing. He also noted the
problems in the current 6bone backbone with sloppy peering arrangements and
poor aggregation.
A major problem for the UK academic networks is that the 6bone backbone
topology is too complex, there is not optimal aggregation, and there are
bad RIPng problems. proposal: get accademic sites to single homed OR use
BGP4+ map of Uk sites:
The graphs from Guy's presentation are available at:
http://www.pipex.net/~guyd/6bone/IETF-presentation/
--------------------------------------------------------
5. Test plan for aggregation-based addressing / Fink
--------------------------------------------------------
Bob Fink noted that the most important work ahead of the 6bone is
conversion to the new Test address format for Aggregation-based Global
Unicast addressing that is about to move to experimental RFC status.
At a meeting later in the IETF week (Thursday at 11:30), those interested
in the 6bone backbone planning met to discuss issues for the
restructuring/cleanup of the backbone, as well as the address conversion.
This meeting was reported in an email to the 6bone mail list that evening,
and is reproduced as an addendum to these minutes.
Then Bob presented the plan for 6bone use of the Test address format, which
is the material presented on the 6bone mail list in late May. This
material is reproduced as an addendum to these minutes.
--------------------------------------------------------
6. Operational issues on the 6bone / various
--------------------------------------------------------
Matt Crawford spoke on multihoming on the 6bone. He noted that a site
multihome to the same provider needs only one prefix if that provider knows
how to handle the extra route.
When connected to two or more providers it will (most likely) be necessary
to have multiple prefixes.
How much interconnection for backbone sites? There is too much as of
today. This is recognized as one topic for discussion as the backbone is
restructured for the new addressing format.
--------------------------------------------------------
7. Implementations in use on the 6bone / Fink
--------------------------------------------------------
Bob Fink reported ion his survey of IPv6 implemenations.
This survey may not be completely accurate, however, it is fairly close
having been circulated for comment on the 6bone mail list.
Host implementations in use on the 6bone are:
Digital OpenVMS Digital Unix
FTP Software Win95 Hitachi NR60
IBM AIX Inria BSD
Linux SICS HP-UX
Sun Solaris UNH for BSD
NRL for BSD WIDE Hydrangea for BSD
WIDE ZETA for BSD WIDE v6d
Host implementations in use on the 6bone are:
Bay Cisco
Digital Fujitsu LR550
Hitachi NR60 Inria BSD
Linux Merit MRT
NRL for BSD Telebit
WIDE Hydrangea for BSD WIDE ZETA for BSD
WIDE v6d
--------------------------------------------------------
Addendum - Report of ad hoc 6bne Backbone planning meeting
--------------------------------------------------------
6bone backbone planning meeting - 14 August 1997, Munich, DE.
Alain Durand held a BOF for those interested in 6bone backbone planning
under the new test aggregation address format. There were 27 people in
attendance.
Alain Durand (G6, FR) spoke on the need to minimize backone tunnels to
clean up routing. There were comments for this, explaing the reasons why
it is needed at this time, and comments as to why we shouldn't worry about
this.
Stephen Stuart (Digital-ca, US) spoke on reasons to cleanup peering, and to
have multiple interconnect points for ISP TLA's.
Matt Crawford showed various multi-prefix scenarios.
There was a general consensus that there was a need to simplify the 6bone
bacbone topology.
Bob Fink (ESnet/LBNL, US) then led a discussion to generate a plan for
readdressing and backbone restructuring. This discussion led to the
following general agreements:
1. that we assign Testing pTLAs (i.e., pseudo TLAs assigned from the NLA1
field of the 6bone Test address allocation) from the Test Aggregation
addressing I-D as follows:
TELEBIT/DK 3FFE:0100::/24
SICS/SE 3FFE:0200::/24
G6/FR 3FFE:0300::/24
JOIN/DE 3FFE:0400::/24
WIDE/JP 3FFE:0500::/24
SURFNET/NL 3FFE:0600::/24
ESNET/US 3FFE:0700::/24
CICNET/US 3FFE:0800::/24
ISI-LAP/US 3FFE:0800::/24
NWNET/US 3FFE:0A00::/24
VIAGENIE/CA 3FFE:0B00::/24
CISCO/US 3FFE:0C00::/24
ANS/US 3FFE:0D00::/24
IFB/UK 3FFE:0E00::/24
NRL/US 3FFE:0F00::/24
CSELT/IT 3FFE:1000::/24
UUNET/UK 3FFE:1100::/24
DIGITAL-CA/US 3FFE:1200::/24
BAY/US 3FFE:1300::/24
UNI-C/DK 3FFE:1400::/24
Note: we started at 1 because Bob is nervous about using 0 :-)
2. that we establish October 1 as the start date for renumbering the
backbone to testing aggregation addresses, with the goal of November 1 for
coming online.
3. that all backbone sites will peer with BGP4+, and only BGP4+.
4. that the old testing addresses (RFC 1897) be discontinued on the
backbone as early as October 1 (by sites already renumbered) and not later
than November 1 when the newly addressed backbone is scheduled to be fully
online.
5. that a call for new pTLA candidates be issued immediately, for inclusion
in the October 1 renumbering/restructuring, where the criteria to be
applied for inclusion is willingness and ability to actively participate in
this timeframe, and demonstrated experience with IPv6 and the 6bone.
6. that a call for existing backbone sites (given a pTLA above) be made to
decide themselves if they are able to participate in this renumbering/
resructuring effort, and be encouraged to give back their pTLA assignment
for now if they aren't able to participate. (Note: any site doing this
can easily reapply at a later time.)
--------------------------------------------------------
Addendum - Call for 6bone Backbone participants email to 6bone list
--------------------------------------------------------
Per the 6bone backbone ad hoc meeting in Munich, I am calling for those
interested in being an early 6bone test pTLA (i.e., pseudo TLAs assigned
from the NLA1 field of the 6bone Test address allocation) when the
renumbering to the new Aggregation-based unicast address format is started
on 1 October.
Requirements are willingness and ability to actively participate in this
timeframe, and demonstrated experience with IPv6 and the 6bone.
Please send your requests to become a 6bone pTLA to the 6bone mail list
with text sufficient to describe your interest and qualifications.
I will assign test pTLAs to all reasonable request at this time.
Thanks,
Bob
-end