Re: [v6ops] IPv6 link-local traffic questions

Hi Toerless,

On Thu, 12 Mar 2020 at 11:01, Toerless Eckert <tte@cs.fau.de> wrote:
>
> Sorry for cross posting, but this is both ops and arch questions.
>
> I was trying to find any generic description that/if link-local IPv6
> traffic MUST/SHOULD use Hop-count 1, but could not find one in
> rfc8200 or rfc4291 and its updates. Instead, text like
> MUST set Hop-limit to 1 is repeated in various RFCs using link-local
> IPv6 traffic.
>

RFC4007  allows packets with link-local destinations to be forwarded
by a router, but only back onto the same link:

"Thus, if a
   router receives a packet with a link-local destination address that
   is not one of the router's own link-local addresses on the arrival
   link, the router is expected to try to forward the packet to the
   destination on that link (subject to successful determination of the
   destination's link-layer address via the Neighbor Discovery protocol
   [9]).  The forwarded packet may be transmitted back through the
   arrival interface, or through any other interface attached to the
   same link."

I think in theory that could also mean forwarded to another router on
the link that then forwards back again onto the link etc., so it is
valid for a packet with a link-local destination address to have an
initial Hop Count value that allows it to be forwarded by multiple
routers, all limited to being forwarded within the same link.

This draft might be useful, as it collects together and summarises
information about using Link-Local Addresses (and is something I've
been meaning to get back to updating, suggestions welcome):

How to use IPv6 Link-Local Addresses in Applications
https://tools.ietf.org/html/draft-smith-ipv6-link-locals-apps-00

Regards,
Mark.

> Am i overlooking something ? Or is this intentional in the IPv6 architecture ?
>
> If intentional, i guess it means: Any RFC introducing some type of link-local
> traffic can specify Hop-limit requirements for its traffic. As in
> "our WG likes 42 as the Hop-limit" (silly to me, but should be fine with
> IPv6 architecture then, right ?).
>
> In one of my draft document reviews using such link-local IPv6 traffic,
> EricV brought up the Q if that traffic shouldn't use RFC5082 instead
> of Hop-Count 1 too. But given how we do not seem to have architectural
> recommendations for TTL=1 (unless, quite likely i overlook something),
> and given how i am not aware of any actual use of GTSM, i am quite
> weary to go down that route. Aka: Would like learn of actual uses of
> GTSM with IPv6.
>
> There is an additional set of consideration re. link-local traffic
> when trying to use it as a HW accelerated tunneling mechanism across links,
> a use we both have in ANIMA, and maybe get the same in BIER (probably
> other WGs too, not sure).
>
> In implementations, i have seen quite a range of issues with
> accelerated forwarding planes across link-local addressed tunnels,
> and when using non-link-local addresses (loopback addresses),
> the HW plane forwarding might have issues doing HW forwarding, because
> usually TTL=1 is only used for control plane traffic. Of course,
> this is not a good architectural argument, but if given architectural
> leeway to even choose 42, maybe GTSM/255 is easier to be supported
> across various HW accelerated forwarding planes than 1.
>
> Opinions and experiences welcome!
>
> Thanks!
>     Toerless
>
> _______________________________________________
> v6ops mailing list
> v6ops@ietf.org
> https://www.ietf.org/mailman/listinfo/v6ops