RIPE 42

RIPE Meeting: 42
Working Group: Routing
Status: Final
Revision Number: 1

Please mail comments/suggestions on:



Draft Minutes Routing WG at RIPE 42
==========

Chair: Joachim Schmitz
Notes: Henk Uijterwaal

A. Preliminaries.
-----------------
The chair called the meeting to order at 9am. Henk Uijterwaal (RIPE NCC)
was appointed as scribe. The minutes from RIPE41 were approved. The agenda
was approved as well. Actions from RIPE41:

* 37.R2: In progress, progress will be reported later.
* 40.R1: Multihoming document: no progress since RIPE 41
* 41.R1: On going, presentation today.
* 41.R2: In progress

B. Routing table analysis/Philip Smith
--------------------------------------
Slides at: http://www.ripe.net/ripe/meetings/archive/ripe-42/presentations/ripe42-routing-bgp/index.html

Philip presented the latest additions to his daily analysis of the
routing table. These include:
- A search of the 192/8 for unallocated/unassigned/unregistered
addresses. Most entries on the list appear to be a matter of
people forgetting to update documentation.
- He fixed count of the number of prefixes smaller than registry
allocations.
- He sees increase in # of prefixes again. The number of prefixes was
constant at about 108,000 between August 2001 and February 2002. Since
February 2002, it is increasing again, now we have 114,000 prefixes.

The chair expressed his worries about the latest increase in the number
of prefixes. This has to be investigated.

No further questions were asked.

C. Identifying unallocated address space.
-----------------------------------------
Slides: not yet available.

C1: BGP unallocated address route server/Philip Smith
This is a proposal from Geoff Huston. The idea is that the RIR's
set up and operate a set of route servers. These server will
provide a BGP feed announcing unallocated addresses. This
proposal has been presented at the APNIC meeting, Geoff is trying
to get consensus on an implementation. Philip mentioned that there
are other ways to do this as well.
C2: Summary of mailing list discussions/Joachim Schmitz.
Joachim summarized the mailing list discussions on this topic. Issues
that were raised there include:
- Unallocated address only, or allocated as well.
- Based on IANA allocations
- Part of RIR efforts.
- Distribution in DNS zone, this was not considered to be a good idea,
as the DNS is already overloaded.
- eBGP feed or simple list
- Include RIS information.
- Administrative: this should be published only as a FYI, not
an authoritative list to avoid legal isseus.
- Life feed vs daily/weekly list
C3: Discussion
Gert Doering: Prefers a list over a BGP feed. He wants to process the
list to filters, a BGP would require him to set up ZEBRA and
extract the list himself.
Bill Woodcock: Agrees and is also worried about losing a BGP update
and thus accidentally filtering addresses.
Philip Smith: What about historic space? Joachim suggests to include
this as much as possible.
Henk Uijterwaal asks if the link to the RIS is considered useful.
C4: Conclusion
Consensus is that this is a good idea. The RIPE NCC (in particular
Andrei Robaschevsky and Henk Uijterwaal) will take this up as an
action.

D. RRCC update/Engin Gunduz
---------------------------
Slides at: http://www/ripe/meetings/archive/ripe-42/presentations/ripe42-routing-rrcc/

Engin gave an update of the RRCC project. A prototype is running and
can be used. He also showed some statistics. Details can be found
in the slides. Plans for the next months include a subscriber service
where a report is mailed at regular intervals and more help on fixing
inconsistencies.

No questions from the floor.

The chair pointed out that this is a useful project, encouraged the
attendees to look at it and provide the NCC with feedback on how to
proceed.

E. IRRToolset update/Katie Petrusha
------------------------------------
Slides at: http://www.ripe.net/ripe/meetings/archive/ripe-42/presentations/ripe42-routing-irrtoolset

The RIPE NCC has taken over support for this package and released a
first update, version 4.7.2. The source as well as pre-compiled binaries
for Solaris, Linux and FreeBSD are available from the RIPE NCC website.
Katie thanked ISI and the mailing-list for their help with finalizing
this release. Future plans include keeping the tools up to date, add
support for multicast and IPv6, and, of course, include suggestions from
the users.

The chair asked if the list of supported platforms could be mentioned
on the web page. This will be done.

A quick show of hands revealed that most attendees had heard of this
product, with a fairly high number of users.

F. RIS update/Matthew Williams-Bywater
--------------------------------------
Slides at: http://www.ripe.net/ripe/meetings/archive/ripe-42/presentations/ripe42-routing-ris/

Matthew gave an update on the RIS. Highlights are: 2 more RRC's have
been installed in Stockholm/Netnod and San Jose/MAE-West. The RIS
architecture has been improved. Several new queries are available, such
as a TOP-N of active prefixes. First results on Black Holes were shown.

Question from the chair: what are other people downloading the data,
doing with the data.

Usage: a show of hands revealed that about 15-20% of the attendees have
used or looked at the RIS pages. Whether this includes the 6 or so
people who looked at the RIS pages from the wavelan network during the
presentation is not clear.

O. RPSLng/Joao Damas
--------------------
Slides at: http://www.ripe.net/ripe/meetings/archive/ripe-42/presentations/ripe42-routing-RPSLng/index.html

Joao gave an overview of expanding RPSL to include IPv6. There is a
pre-Internet draft on this topic, released last Friday (26/4). The
draft is being discussed on the rpslng _at_ ripe _dot_ net list. It will be
submitted to the IETF, the RPSL IETF WG will not be reopened for the
time being. Joao presented the key ideas of the draft.

Questions/Discussion:
The chair was happy with the progress made but warned that there still
is a lot of work to be done, both in the standard as well as for
deployment. IRRtoolset will need to be updated, though Katie Pretushka
is looking at it.

Ping Lu (C&W): How do we specify mandatory and optional attributes
Joao: Solve by having mandatory attributes "afi" otherwise parser
will flag this as an error
PL: But what about strictly IPv4 code?
Joao: Attribute forces state

Chair: Can we incorporate sub-attributes?
Joao: I have to think about this.

Randy Bush likes to separate standards issues from implementation
and deployment issues. Do standard without code being there.
Joao: agrees with this comment.

X. Meeting closed @ 11:30