Re: [16NG] Re: multicast and IPv6 over ETHCS

[Alexandru Petrescu <[e-mail]>]

David Johnston wrote:
> I don’t agree with that.

David, I'm trying to identify whether your disagreement matches my
understanding.

What don't you agree with?  You don't agree that IP-over-ETHCS
document should not specify how the link-layer implements link-layer
multicast?  In other words do you suggest that the document _should_
specify how the link-layer implements link-layer multicast?

> An UL Ethernet MSDU with a multicast or broadcast DA should get sent 
> to members of the multicast group/everyone in the bridged LAN.

Yes.  Is this a problem?

> Since 802.16 implements a point to point Ethernet similarly to modern
>  cat 5 ethernet, then either everyone=the Ethernet termination in the
>  BS, or it hits an 802.1D bridge in the BS and so the bridge deals 
> with sending it to the other 802 LANs to which is it has bridge 
> ports, including other SSs.

I sense this is a way of implementing it.  It is however a link-layer
method.

> As I mentioned before, an 802.16 BS does not implement one network, 
> it implements a bunch of point to point networks that may be joined 
> by a bridge or IP routing or any other scheme that is implemented in 
> the BS.

What is that IP routing scheme that implements a link-layer?  Is it
MPLS?  What other?

Do you suggest that the draft IP-over-ETHCS recommends the 802.16 BS to
implement a bunch of point to point networks joined by an IP router?

I think the draft _could_ list some means by which the link-layer
multicast  _could_ be implemented (major use of word 'MAY') but not to
make any recommendation about that.

Alex

> 
> 
> DJ
> 
> 
> 
> 
> 
> 
> 
> ------------------------------------------------------------------------
> 
> 
> 
> 
> 
> *From:* Jihoon Lee [mailto:[e-mail]] *Sent:* Monday, 
> January 15, 2007 4:12 PM *To:* Alexandru Petrescu *Cc:* 
> [e-mail]; [e-mail] *Subject:* [16NG] Re: multicast and
>  IPv6 over ETHCS
> 
> 
> 
> Hi Alex,
> 
> I agree with that.
> 
> 
> 
> Thanks,
> 
> Jihoon
> 
> 
> 
> 
> 
> 
> 
> 2007/1/15, Alexandru Petrescu <[e-mail] 
> <mailto:[e-mail]> >:
> 
> Hi Jihoon,
> 
> Jihoon Lee wrote:
>> Hi Alex,
>> 
>> Sorry for my late response. Basically I agree with your opinion. An
>>  IPv6 node requires a link local multicast address in order to 
>> perform DAD, ND, and RA.
>> 
>> Contrary to our expectations, 802.16/16e don't support UL multicast
>>  (which means an MS forwards data directly to other MSs in the same
>>  cell).
>> 
>> (BTW, I think I need to clarify the UL multicast I mentioned. I 
>> believe you already know this: the multicast/broadcast which is 
>> sent by an MS  to other MSs in the same cell (BS) cannot be 
>> supported in 802.16. However, other MSs in another cell (BS) may 
>> receive this since the 802.16 backhaul (WiMAX ASN) may 
>> multicast/broadcast at ETH or IP layer.)
>> 
>> I agree that the document needs to specify how to deal with this 
>> problem.
> 
> Yes, I think the IPv6-over-ETHCS document should make clear the 
> assumption of IPv6 using link-layer multicast for IPv6.  I don't 
> think the IPv6-over-ETHCS document should solve the problem you 
> mention above between parenthesis (uplink mc, or mc between BSs) at 
> link layer.
> 
> The IPv6-over-ETHCS document could list the behaviour of ETHCS it 
> expects.  Ie:
> 
> -when MS sends NS to a IP solicited-node multicast address the node 
> who has joined group must receive it. -when BS or AR advertises an RA
>  to all-nodes multicast address then all SSs who have joined that 
> group must receive it. -mapping an IPv6 link-local multicast address 
> should happen (eg sending a packet to IP dst ff02::1 should have the 
> link-layer dst 33:33:33:0:0:0:1).
> 
> The document IPv6-over-ETHCS should list what are the means for SS to
>  _join_ a link-layer multicast group.  There are few of them (send a
>  link-layer message, send a MLD message, establish a local filtering
>  rule, always receiving all packets, etc).
> 
> But I don't think the IPv6-over-ETHCS document should specify the 
> link-layer behaviour for an implementation of ETHCS link-layer 
> multicast.  Do you agree with this latter item?
> 
> Alex
> 
>> 
>> Best regards, Jihoon
>> 
>> 
>> 2007/1/12, Alexandru Petrescu < [e-mail]
> <mailto:[e-mail]>
>> <mailto:[e-mail]
> <mailto:[e-mail]>>>:
>> 
>> Hi Jihoon, thanks for reply,
>> 
>> Jihoon Lee wrote:
>>> Hi Alex,
>>> 
>>> Let me jump into discussion. 1) 802.16/16e MAC has no capability 
>>> to do uplink multicast. In DL, 802.16 provides multicast CIDs 
>>> which is initiated by DSA messages. In UL, however, there is no 
>>> way for an MS to access others' UL data fundamentally, in 802.16 
>>> PHY/MAC.
>> 
>> Ok.  For one, if the 802.16MAC is not capable to do bidirectional 
>> (or normal) multicast then that is a big issue for 
>> IPv6-over-IPv6CS. (and surprisingly, apparently the document 
>> IPv6-over-IPv6CS doesn't seem to mention the word 'multicast'.)
>> 
>> The issue is that a SS running IPv6 needs to multicast a NA, from 
>> time to time.  For DAD it needs to send a NS to a multicast address
>>  too.
>> 
>>> In case of ETH over 802.16, the ETH(bridge) may cover this (an MS
>>>  sends multicast data in UL, and then a bridge forwards it back).
>>>  But, there is still a difficulty in multicasting data back 
>>> except for the source MS.
>> 
>> You mean the bridge in BS?  I was thinking ETHCS in SS may offer a
>>  multicast interface to the IP stack.
>> 
>> In both cases (bridge in BS or bridge in SS), I think it is not up 
>> to this document to specify how the ETHCS transforms a asymmetric 
>> ul/dl multicast feature into a symmetric one.  But it should be a 
>> goal for ETHCS to offer such an interface to the IPv6 stack, 
>> otherwise the IPv6 stack won't work.  What do you think?
>> 
>> Alex
>> 
>> 
> 
> 
> 


_______________________________________________
16NG mailing list
[e-mail]
https://www1.ietf.org/mailman/listinfo/16ng