[6bone] free transit

[Pim van Pelt]

On Tue, Dec 03, 2002 at 03:11:38PM +0000, Hans Goes wrote:
| On Tue, 3 Dec 2002, Jeroen Massar wrote:
| 
| > Indeed, unless you have a big worldwide IPv6 network like some lucky
| > people have.
| > You might want to check up on:
| > http://ip6.de.easynet.net/ipv6-minimum-peering.txt
| > and
| > http://www.ietf.org/internet-drafts/draft-savola-v6ops-6bone-mess-01.txt
| 
| 
| Ok.... But do we accept this way of routing ? I think it's a bad thing.
You should not, in my opinion. One thing you can do, is filter inboud
announcements to fit into this regexp:
^([0-9]+)$

This will accept paths of exactly one AS, effectively eliminating all
the stuff you get from your peers. If you don't want to have all the
prefixes, the best thing you can do is request that your peer only send
their and their customers routes (normally, this would be only one prefix).

If they do not honor your request, I think you should remove peering. By
the way, at AMS-IX (where you recently migrated from 1890 to 12702
also), I don't see heavy transit either. Almost all of my peers send me
only their own prefixes. I myself send all tunneled peers only the
prefixes I learn at AMS-IX.

I think it's also time to stop building far-away tunnels. A rule of
thumb might be: only tunnel to peering ASes. Making a tunnel from
Amsterdam to California via 2 or 3 transit ASes is not what you want.
The traffic will get there sooner or later without you (note: not you
personally :) making an explicit tunnel to them.

We might also want to start accepting only 2001:400::/23 prefixes from
our ARIN peers, and 2001:200::/23 from our APNIC. This way, you cannot
'go to europe via america' in 2001::/16 land. I'm sorry to say that this
will not hold for 6BONE, as these prefixes are allocated sequentially
and not from supernets.

Hope this helps.

groet,
Pim
-- 
---------- - -    - - -+- - -    - - ----------
Pim van Pelt                 Email: pim@ipng.nl
http://www.ipng.nl/             IPv6 Deployment
-----------------------------------------------