Re: [v6ops] Filtering anycast 6to4 [draft-ietf-v6ops-6to4-to-historic WGLC]

[Ray Hunter <[e-mail]>]

Tore Anderson wrote:
> * Ray Hunter
>
>> The new BCP text that I proposed is a SHOULD NOT _advertise_.
>>
>> If your AS chooses to _learn_ the route to 192.88.99.1 from others
>> for your own operational reasons, that's entirely up to you, and I've
>> proposed no BCP for that.
>
> A transit AS who learns the 6to4 prefixes from an upstream or peer
> will advertise it to its customers. I initially intepreted your
> suggested text to understand that a transit AS should explicitly
> suppress such advertisements to my customers, but from the above I
> understand that is probably not the intention? If so, perhaps replace
> "advertise" with "originate"?
>
> Tore
>
>
After reading the various reactions, your amendment would seem to result 
in a reasonable piece of BCP that everyone could live with, allowing 
graceful sunsetting, whilst avoiding "bad" relays.

So I propose:

s/Internet service providers SHOULD filter out routes to 
192.88.99.1./Internet service providers and other network providers 
SHOULD NOT originate a route to 192.88.99.1, unless they actively 
maintain a path to an outbound 6to4 relay service as detailed in 
http://tools.ietf.org/html/rfc6343 Section 4.2.1/


-- 
Regards,
RayH