[6bone] Dualing IPv6 announcements

[Jeroen Massar]

Hank Nussbacher wrote:

> BT and JPNIC/WIDE are both announcing these prefixes:

Did you contact them directly?
 
> mcast#sho bgp ipv6 inc

Check
http://www.sixxs.net/tools/grh/lg/?prefix=2001:200::/32&matchtype=more
or
http://www.sixxs.net/tools/grh/lg/?prefix=2001:200::/32&matchtype=more&s
how=origins
to show just the origins of these prefixes (for lazy people like me ;)

Which boils down to:
2001:200::/35        1752 BT-CIN-AND-ADASTRAL 
2001:200::/35        2500 JPNIC-ASBLOCK-AP 
2001:200::/32        1752 BT-CIN-AND-ADASTRAL 
2001:200::/32        2500 JPNIC-ASBLOCK-AP 
2001:200:126::/48   10017 JPNIC-NET-JP-AS-BLOCK 
2001:200:200::/40    4682 AS4682 
2001:200:202::/64   10017 JPNIC-NET-JP-AS-BLOCK 
2001:200:202::/48   10017 JPNIC-NET-JP-AS-BLOCK 
2001:200:202:f::/64 10017 JPNIC-NET-JP-AS-BLOCK 
2001:200:203::/48    9377 AS9377 
2001:200:340::/48    9377 AS9377 
2001:200:500::/40    9612 JPNIC-NET-JP-AS-BLOCK 

> so the question is why is BT announcing this and how come 
> prefix filters aren't stopping this?

Because not very many ASN's apply filtering *at all*.
Let alone filter something that could, even how odd, be legit.

The two /64's in the above table should never be seen
as 10017 could quite easily aggregate them into the /48
and even then... the /48 should not pop up globally.

2001:200:202::/48   > 3ffe:81d0:ffff:2::28 1000   6939 14277 3549 2915
2713 4682 10017  IGP 
2001:200:202::/64     3ffe:81d0:ffff:2::28 1000   6939  6939 3549 2915
2713 4682 10017  unknown 
2001:200:202:f::/64   3ffe:81d0:ffff:2::28 1000   6939  6939 3549 2915
2713 4682 10017  unknown

This just shows how badly aggregation and filtering is done in the IPv6
world.
Also note that the /64's are marked as coming from 'unknown' instead of
IGP.

So who is to blame here? Nobody, you are allowed to set up your own
policies.
It's your own network and if you care for it or not, alas...
The bad part is that if everybody starts announcing (and not
aggregating) their
announcements, the routing table will explode.

"Admins" should be reading and applying:
Minimal IPv6 Peering (*1) by Robert Kießling 
Moving from 6bone to IPv6 Internet (*2) by Pekka Savola.

Also I think that it is much worse that people don't even notice
that their own address space is announced doubly/oddly etc or that
it is ghosted without their knowledge.

Greets,
 Jeroen

*1 = http://ip6.de.easynet.net/ipv6-minimum-peering.txt
*2 =
http://www.ietf.org/internet-drafts/draft-savola-v6ops-6bone-mess-01.txt