Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: [EFM] (EFM)a question about P2P




Hi, Dolors:
Maybe I didn't understand your thought well and also have some misunderstand of 802.1D. 
in the email your say it is a shared emulation, I understand that you want to solve both the shared media and point to point link, I think it's better if we can meet these two requirements. do you agree this?

about the shared media I still have some questions.
in the shared media, ONU must detect and discard own frames in downstream direction, 
but I don't know ONU how to treat the  frames that their destination address is within another ONU, in this
ONU's MAC address table, there is no this MAC entry, so it'll forward this frame to all the other ports belong to this ONU. do you agree this or have some other choice?
maybe we are back to the first email disscussion about the security problems in share emulation.

thanks for you kindly reply


xu zhang



>Xu Zhang,
>
>I am having trouble understanding your sentences. I think the presentations we have show that it is possible. We are not presenting a p2p layer on top of the MAC. It is a shared emulation. All options are possible.
>
>I am not sure I understand what you are suggesting. Can you point to specifics on the presentation?
>
>Dolors
>
>
>zhangxu wrote:
>
>> Dolors:
>>  Thanks for your reply,  I think I didn't explain my  thought well,
>> now I'll explain why I think the p2p will be located under the MAC layer.
>>  if the p2p was located above the MAC layer, so the MAC is transparent to the p2p layer, the 802.1D relay function located between two MAC layer, so the p2p was located above the relay or bridge function, it can't solve the EPON compatible problem with 802.1D.
>>
>>
>>                     xu zhang
>>
>> >Xu Zhang thanks for your comments,
>> >
>> >You are presenting an interesting security issue that we should
>> >consider. However, I don't see how it relates to the decision of
>> >p2p-emulation below MAC versus forwarding rules above the MAC.
>> >
>> >A shared-emulation model sends just one copy of this packet in the
>> >broadcast medium. The decision is what encryption key to use since
>> >the destination is not known yet.
>> >
>> >The p2p-emulation sends as many copies as virtual links. It can use the
>> >key of the destination (since there is only one for each copy) in each
>> >virtual link. However, this still means that all ONUs can see the copy
>> >of the frame, but only one is the actual destination.
>> >
>> >So both cases seem to be equivalent from this point of view.
>> >
>> >Dolors
>> >
>> >zhangxu wrote:
>> >
>> >> Hi, Dolors:
>> >>
>> >> thanks for your september presentation, I am very interested about it. I think whether the P2P layer be put above the
>> >> MAC layer or under the MAC layer is a very important question.
>> >> I think if we put the P2P layer about the MAC layer then using filter layer to emulate P2P between different ONU,
>> >> this can solve the problem that without PTP layer different OUN can't communicate each other in layer 2,
>> >> but there is a question that we know downstream in PON is broadcasting nature, the ONU how to deal with the ethernet frame with unknown destination addr if the destination addr is an addr in another ONU range.
>> >> traditional ethernet switch will forward the frame to all the other ports(because the destinaton addr is unknown), I think in EPON if we solve the problem like that is not suitable for the security problem, we must filter these packet out.
>> >> where we can filter these packet out, I think under the MAC layer is more suitable,
>> >> so I prefer put the P2P layer under the MAC layer.
>> >>
>> >> best regards
>> >>
>> >>                    xu zhang
>> >>                    bu. tele. tech. Co, Ltd
>> >>                    Email: zhangx@xxxxxxxx
>>