Re: Generic anycast addresses...

On Thu, May 30, 2019 at 10:30:12AM -0400, Michael Richardson wrote:
> RFC7084 points out:
> 
>    Anycast addresses are syntactically indistinguishable from unicast
>    addresses.  Anycast addressing is equivalent to that of unicast in
>    multiple locations.
> 
> and I think you want an address which is synatically distinguishable from
> unicast.  Maybe something that does not get caught by default route; although
> I'm not sure about that.

+1

If we want to improve support for anycast, i would start figuring out
what we could do in the routing protocol domain. I am not aware of
any explicit support for anycast thats actually widely deployed.
We are still running anycast routing (for decades now) as an unexpected
bonus from the way our routing protocols work.

Indeed, it may be more difficult right now to operationally discover
what Mark called the "fault", e.g.: discovering when the same prefix
is incorrectly announced from two different originating sources than
just assuming its path redundancy or anycast. Appropriate tagging
sourced route information could help a lot to improve diagnostics
and allow more flexible anycast policies. main issue is tht routing
folks don't like to do new stuff unless their pain level is high enoug,
and "anycast routing kinda works well enough right now" (famous last words).

In any case i am a lot more a fan of solving the policies of addresses
being anycast rather than "fault" in the routing than the RIR domain.

toerless

> Maybe we could have hacked in what you want into site-local space, but that's deprecated.
> I don't think it can go into link-local space.
> 
> In the case where the service is not available in the homenet/campus, where
> would the packet get discarded?   Maybe this could be done in an IANA
> allocated /48 that we all agree to blackhole route on the DFZ, but this seems
> like it might be a hack.
> 
> --
> Michael Richardson <mcr+IETF@sandelman.ca>, Sandelman Software Works
>  -= IPv6 IoT consulting =-
> 
> 
> 

> --------------------------------------------------------------------
> IETF IPv6 working group mailing list
> ipv6@ietf.org
> Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
> --------------------------------------------------------------------

-- 
---
tte@cs.fau.de