Re: Metadata over IPv6

[[e-mail]]

Hi Brian,

>> Thanks for the question.
>> What comes quickly to mind is RFC 1546, which describes a local "Host Anycasting Service". Related, https://tools.ietf.org/html/rfc4291#section-2.6 and https://tools.ietf.org/html/rfc4291#section-2.7, which respectively talk about local anycast and scoped multicast (one of the scopes being "local").
>> The bit pattern of a link-local address might be useful. That said, I would want this to not be "one-of", known by a few that happen to use a proprietary service (AWS), but consistent with the IPv6 addressing architecture and openly documented. From that perspective, I should think I might want a short RFC (I can help with drafting, if you like) that identifies the bit pattern and talks about it a bit.
>> When documented in that way, one could imagine IANA picking it up. They do that.
> 
> Hi Fred,
> 
> So it seems you and Mark are of the same mindset that I should request a new anycast allocation for this, which I'm fine with because it becomes official, even if it does take more time.  It should be a very short draft.
> 
> And just for my own sanity, this anycast address would apply to an address of any scope, correct?  For example, given 7c is the next one available:
> 
> fe80::7c
> 2001:db8::7c
> 
> I just want to make sure I cover the case where there isn't any configured IPv6 address besides the link-local.

Is you problem solvable using "service discovery" (e.g. DNS-SD)?

Best regards,
Ole