Re: [v6ops] I-D Action: draft-colitti-v6ops-host-addr-availability-01.txt

[Mark Smith <[e-mail]>]

Hi,

On 24 July 2015 at 17:45, Brian E Carpenter <[e-mail]> wrote:
> I like -01 even better than -00.
>

So do I.

> As well as the reference [TARP], you might add draft-smith-enhance-vne-with-ipv6
> as a use case for a /64 per host (specifically per NVE, network virtual edge).
>

I've never really thought of tunnel end-points (NVEs in that draft) as
'hosts', however I suppose they are if they're the final destination
for the packet, as NVEs are for underlay network packets.

I've been thinking a bit recently about how to describe/determine
where the network "finishes" and the host/end "starts", partly in the
context of things like Segment Routing and "service function chaining"
or "network function virtualisation". I've thought the simple and
fundamental rule is that the network ends and the host/end starts at
the "payload don't care/care" boundary. If the receiving device
doesn't care what the payload is (and therefore doesn't care if the
payload is encrypted or not), then it is a device in the network. If
receiving device cares about what the payload is, then it is
performing host/end functions.

By that definition, even though NVEs are only performing layer 2/layer
3 frame/packet processing, they're hosts/ends, because they look at
and process the payload of the packet they receive that are directed
specifically at them (i.e., packet dest addr is their own /64 prefix
in that draft).

Regards,
Mark.


> Regards
>    Brian
>
> _______________________________________________
> v6ops mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/v6ops