Re: [ipwave] WGLC for draft-ietf-ipwave-ipv6-over-80211ocb-08

[Tony Li <[e-mail]>]

Hi,

> [AB] Why will WG choose only the Ethernet framing in WAVE? or why we need to force all IP packets to be encapsulated by Ethernet? Why not an adaptation for one general framing related to IPWAVE applications and enviroments?
> 
> [AB] The answer for the above seems to me like, it is easier for design, but is it efficient for WAVE new generations?
> 
> [AB]  If we say let us look into bridging wireless and wired (which is not this document objective), we can have bridging Ethernet and WiFi, but is that most available WAVE protocols in the wired networks?  



If you haven’t noticed, the document doesn’t talk about using IP over WAVE. That’s already been defined elsewhere and isn’t relevant to our work. The goal for this document is to establish and document framing for IP over 802.11-OCB. 


> Once you ignore all of the (albeit extremely important) L2 mechanisms that go on behind the scenes, 802.11 looks just like an Ethernet interface. If we do it right, 802.11-OCB will too. That’s the whole point of this document.
> 
> So is the point to ignore that we have a wireless network, so we assume our network is wired. That was many old approaches in IPv4 networks, we need to move on to better approaches for IPv6. This sounds like we assume what is not true to be true, so that means BAD service. IMHO, those assumptions in the old days of IPv4 and now in many IPv4 networks can work ok because they are fixed-wireless, they have no WAVE scenarios, now I think our WG are required to look into transmitting IP packets into real scenarios of WAVE.
> 
> [AB] Why will we ignore L2 mechanism (which is 802.11 in our doc title as an objective) but still the main objective is to transmit over it? I think we need to be clear in the document of such assumption or ignorance.
> However, I am not in favour to ignore the 802.11 mechanisms just to make things simple.


That’s a bold claim, and one that I can’t agree with. My work with DSRC shows that it works just fine without any special considerations. To the host, it looks like an Ethernet. If you have evidence to the contrary, please put it on the table. Until then, please stop spreading Fear, Uncertainty and Doubt.

We will ignore the L2 mechanisms because they are out of scope. It is not in the IETF’s purview to document or change L2 mechanisms. We can, of course, use documented mechanisms.


>> [AB] How can it be used by IPv6?
> 
> What does “it” refer to?
> 
> The layer of adaptation mentioned in the doc.


Very simple: it allows the host layer to see a simple Ethernet encapsulation. It’s a wee bit of complexity in the device driver, to be sure, but above that layer, the rest of the host just sees Ethernet.

I did this for my implementation and ARP and IPv4 just worked. It makes implementation trivial.

Tony