RIPE 44

RIPE Meeting: 44
Working Group: Routing
Status: Final
Revision Number: 2


RIPE 44, Amsterdam
Routing Working Group Session
28-January-2002 Minutes

Chair: Joachim Schmitz - JS395-RIPE
Co-chair: Joao Damas - joao _at_ psg _dot_ com
Scribe: Arife Vural - AC10172-RIPE -- thank you!!
Participants: 99


A. Preliminaries

Joachim Schmitz welcomed us all to the meeting and declared it open. The
Participant's list was then handed out by the chair and Arife Vural from
RIPE-NCC volunteered to take the minutes.

Joachim Schmitz mentioned the overlap between Routing and DNS and IPv6. He
will try to avoid overlaps but it may not be ruled out entirely. During the
second slot of Routing WG there will be some issues discussed of IPv6
routing.

The minutes of the previous meeting at RIPE43 were approved.

There is no change to the agenda which was published before RIPE44.


B. Previous meetings actions

40.R1 Creating multi-homing doc is in progress Philip Smith and Joachim Schmitz
had been discussing this offline. Joachim Schmitz suggested to drop this item
until they may come up with something. Philip Smith thinks everything is
ready about the doc but he could not find the time to write them down.

41.R1 Work on RPSLng, M.Blanchet, D.Kessens, F.Parent, J.Schmitz, RIPE NCC.

41.R2 Work on RRCC, RIPE NCC

42.R1 Implement objects in database for unallocated IP address space, RIPE
NCC

42.R2 Arrange with other registries coverage of complete unallocated IP address
space, RIPE NCC.

42.R3 Identify unallocated IP address space in RIS, RIS team.


C. Co-chair of Routing-WG

Joachim Schmitz said that Joao had already acted as an interim chairman when
Joachim could not attend the meeting and he had done a good job. Also, he has
been instrumental in several projects in the past related to the work of the
Routing WG as the RRCC project. So, Joachim proposed Joao be co-chair of
Routing-WG. Joao was accepted. After this assignment, Joachim is looking
for another co-chair in case of absence of Joao and Joachim. If anyone has a
candidate, let Joachim or Joao know.


D. Allocated Space, Status and Next Steps, Shane Kerr.

Slides are accessible from

http://www.ripe.net/ripe/meetings/archive/ripe-44/presentations/ripe44-routing-allocated/

Shane explained what their motivation is to publish allocated address space in
the RIPE region. He said that he would like to help people to update their
filters by using this data. As a first step, obviously the internal and whois
DB must be made consistent.

Joachim Schmitz suggested to go and verify the unallocated address space and
then have seen how much of the unallocated space is announced.

Results get from

ftp://ftp.ripe.net/ripe/stats/issued/

Beyond Allocated Space, Shane gave a quick overview about the RRCC project which
provides user to find out the inconsistencies between whois DB and routing
tables. Now, user can get an e-mail report in case of inconsistencies. Source
code of the tool is open to everyone and can be downloaded from

ftp://ftp.ripe.net/ripe/dbase/software/RRCC-0.2.tar.gz


E. Randy was not in Amsterdam and his presentation was cancelled.


F. Internet Routing Table Analysis Update, Philip Smith.

http://www.ripe.net/ripe/meetings/archive/ripe-44/presentations/ripe44-routing-bgptable.pdf

Philip Smith presented the latest changes and his observation of the Weekly
Routing Table Report. He added LACNIC into the report. He pointed out the very
high number of prefixes /24:65807

CIDR Report (second part of Philip Smiths' presentation)

During the last couple of months, the increase in the number of aggregates
continues and growing of the BGP Routing Table is slow. 2002 looks better in
aggregation and growing is slow but it looks as if the trend in 2003 develops
more towards earlier steep increases seen. The reasons behind that are still
unknown, but the purpose of the CIDR report is diagnosis rather than
analysis.

Christian Panigl: When you look through PS reports, RIPE region is doing a
good aggregation.


G. Routing Information Services, Status and Plans, Matthew Williams.

Slides at
http://www.ripe.net/ripe/meetings/archive/ripe-44/presentations/ripe44-routing-ris/

Matthew Williams presented the latest status of RIS Project. The RIS Project is
looking for a new location in Mea-East and Middle-East. There are new
developments like RIS BGP Beacons, RISng, PrefixInuse utility. Also there are
some ongoing projects. MyAS is one of them which allows network operators to
view their AS number in global routing tables. Beyond that, Matthew
presented how DDos attacks on DNS root servers is seen by RIS.


H. RISng - Technical Overview, James Aldridge.

Slides at

http://www.ripe.net/ripe/meetings/archive/ripe-44/presentations/ripe44-routing-risng/

James Aldridge gave a technical overview of the new generation of RIS DB.
Running RIS DB has some bottlenecks in central server structure. To sort it out
a distributed service structure had been developed.

Also James Aldridge presented some graphs which showed how the ms-SQL worm
became visible in routing tables and how it is observed by RIS.

Slides at
http://www.ripe.net/ripe/meetings/archive/ripe-44/presentations/ripe44-routing-ms-sql.pdf


I. BGP Beacon Data and Other Statistics, Henk Uijterwaal.

Slides at

http://www.ripe.net/ripe/meetings/archive/ripe-44/presentations/ripe44-routing-beacon/

Henk Uijterwaal presented the analysis he has done by using RIS BGP Beacon
data. Also, he presented the statistics about AS numbers, and Prefixes. One of
the interesting point in his presentation, one out of 3 ASes is never seen on
routing tables.

Ouestions? Comments.

Christian Panigl: How did Henk produce those numbers like single-home or
multi-homed by looking at the data only IX points? At that point another
question comes up, how much of the Internet is seen by RIS?

Henk Uijterwaal: This is difficult to tell, comparing data with other Internet
studies and with assigned address space indicate a certain level of
completeness. Much of the global AS topology should be captured, but definitely
not the majority of local configurations - after all RIS tests only at certain
sample points in the topology (definition of local?)

Christian Panigl: What happens if a prefix is re-announced before withdrawal
has finished propagating?
This could be tested with RIPE RIS Beacons to find out why route-flap damping
does not work as expected.

Christian Panigl: Is it possible to provide data which architectures do many withdrawals?
It may be deducted to a certain extent. Regarding version of router operating
systems this is not necessarily obvious.

Wilfried Woeber: RIS does not see the entire picture and because of that it is
not right to make generalization.

Joachim Schmitz proposed to find out how much of the Internet is seen by the
RIS. Also he suggested to be more careful when phrasing statements about the
status of the Internet, or drawing conclusions on a general level.

Christian Panigl said that RIS is doing a great job.


J. RPSLng, the new draft, Joao Luis Silva Damas.

Slides at http://www.ripe.net/ripe/meetings/archive/ripe-43/presentations/ripe43-routing-rpslng/.

Joao talked about RPSLng Internet Draft he published last Xmas. There has been
some discussions about it recently on the mailing lists. The intention was to
publish the second version of the draft before the RIPE meeting but as it seems
discussion was picking up recently. Joao thinks it's better to publish it after
the RIPE Meeting in two weeks time. At the same time, the RIPE NCC is doing
preliminary work on implementing the draft.
It is very important to try to implement it before moving on as the
implementation effort would show possible problems in the draft.

Larry Blunk from Merit reports that their implementation is almost ready as well.
Finally, Joao reports that he submitted the draft as personal draft but suggests
that successive ones may be submitted from the IETF RP3 WG.


K. Operational Feedback to IP Equipment Vendors, V.Gill.

Slides at http://www.vijaygill.com/nanog26/feedback.ppt

This talk is given by Joachim Schmitz rather than his colleague V. Gill.
It can obviously not be avoided that WGs are running in parallel. Joachim
explains that he knows about substantial overlap with the Database, IPv6,
and EIX WGs. Joachim asked about who sees an impact with other WGs, and it
turns out that the overlap with DNS (counting hands) is also not negligible.
Joachim will work for the next meeting to minimize the overlap.
If anyone has any comment or suggestions, please contact Joachim Schmitz.