RE: [16NG] Fw:I-DACTION:draft-madanapalli-16ng-ipv4-over-802-dot-16-ipcs-00.txt

["Narvaez, Paolo \(Paolo\)" <[e-mail]>]

 

> > If there is concern for a legacy IPv4 stack at the MS sending ARP 
> > requests and perhaps even requiring ARP responses, why not 
> recommend a 
> > shim layer between the IPv4 stack and the 802.16 IPCS layer 
> at the MS. 
> > This layer would filter ARP requests arriving from the IPv4 
> stack and 
> > respond with any MAC address. The IPCS classifier should 
> never see ARP 
> > messages, and they should never be sent over the air.
> 
> This can be a solution. But can we leave this for 
> implementors? And in the draft we just say ARP is not 
> required and MUST not be sent over the air.


I agree. There is no need to provide details of how the MS deals with
this issue. However, we cannot forbid ARPs over the air because ARP
packets over the air (IPCS) do not exist nor are they defined. It might
be more clear to say that the MS should not send any ARP-related
messages over the air.

Paolo

> -----Original Message-----
> From: Syam Madanapalli [mailto:[e-mail]] 
> Sent: Wednesday, February 14, 2007 2:58 AM
> To: Narvaez, Paolo (Paolo)
> Cc: Behcet Sarikaya; [e-mail]
> Subject: Re: [16NG] 
> Fw:I-DACTION:draft-madanapalli-16ng-ipv4-over-802-dot-16-ipcs-00.txt
> 
> Hi Bachet, Paolo,
> 
> In the next revision I will add an explicit section for the 
> subnet model.
> Please some inline text.
> 
> On 2/13/07, Narvaez, Paolo (Paolo) <[e-mail]> wrote:
> >
> >
> >
> > How can the MS and AR send ARP requests and responses to 
> each other? 
> > These ARP messages have to go over IPCS and there is no IP 
> packet in this case.
> 
> I agree with this, as there is no IP header, IPCS cannot 
> classify these frames.
> 
> >
> > If there is concern for a legacy IPv4 stack at the MS sending ARP 
> > requests and perhaps even requiring ARP responses, why not 
> recommend a 
> > shim layer between the IPv4 stack and the 802.16 IPCS layer 
> at the MS. 
> > This layer would filter ARP requests arriving from the IPv4 
> stack and 
> > respond with any MAC address. The IPCS classifier should 
> never see ARP 
> > messages, and they should never be sent over the air.
> 
> This can be a solution. But can we leave this for 
> implementors? And in the draft we just say ARP is not 
> required and MUST not be sent over the air.
> 
> 
> Thanks,
> Syam
> 
> >
> > Regards,
> > Paolo
> >
> >
> >  ________________________________
> >  From: Behcet Sarikaya [mailto:[e-mail]]
> > Sent: Monday, February 12, 2007 2:19 PM
> > To: Syam Madanapalli
> > Cc: [e-mail]
> >
> > Subject: Re: [16NG]
> > Fw:I-DACTION:draft-madanapalli-16ng-ipv4-over-802-dot-16-ipcs-00.txt
> >
> >
> >
> >
> > Hi Syam,
> >   You mentioned P2P link here in this message but I could 
> not see any 
> > text on this in the draft. Maybe a section should be added on the 
> > address and subnet mask assignment given the P2P nature of the link 
> > and the fact that such assignments are not commonly used.
> >   You recommended AR to ignore ARP Requests from MS, I am 
> not sure if 
> > this is OK or a Reply would be needed according to ARP?
> >
> > Regards,
> >
> > Behcet
> >
> >
> > ----- Original Message ----
> > From: Syam Madanapalli <[e-mail]>
> > To: "Narvaez, Paolo (Paolo)" <[e-mail]>
> > Cc: [e-mail]
> > Sent: Saturday, February 10, 2007 12:34:46 AM
> > Subject: Re: [16NG] Fw:
> > I-DACTION:draft-madanapalli-16ng-ipv4-over-802-dot-16-ipcs-00.txt
> >
> >
> > Hi Paolo,
> >
> > On 2/10/07, Narvaez, Paolo (Paolo) <[e-mail]> wrote:
> > >
> > > I have two general comments on this draft.
> > >
> > > 1. Section 7 assumes that ARP messages sent by the MS are 
> received 
> > > by the AR. The section states that ARP messages should be 
> dropped by 
> > > the AR. However, since ARP runs directly over Ethernet 
> (there is no 
> > > IP
> > > header) how can the 802.16 IPCS carry this type of packet 
> over the air?
> >
> > As mentioned the draft, ARP is not required for 802.16 
> because of the 
> > fact that the MAC address is not required for identifying 
> the destination.
> > There was a concern that an unmodified host may send ARP without 
> > knowing that the underlying interface is 802.16, because of this I 
> > added the above statement.
> >
> > If we think there is no such possibility, then I will 
> remove this text.
> >
> > >
> > > 2. What is the subnet assignment? Can two MSs be on the 
> same subnet? 
> > > Or do we need to introduce some restrictions as in the 
> > > IPv6-over-802.16 draft.
> >
> > The draft assumes the p2p link, so no two MSs can be on the 
> same subnet.
> >
> > Thanks,
> > Syam
> >
> > >
> > > Regards,
> > > Paolo
> > >
> > >
> > >
> > >
> > > > -----Original Message-----
> > > > From: Daniel Park [mailto:[e-mail]]
> > > > Sent: Thursday, February 08, 2007 7:17 PM
> > > > To: [e-mail]
> > > > Subject: [16NG] Fw:
> > > >
> > I-DACTION:draft-madanapalli-16ng-ipv4-over-802-dot-16-ipcs-00.txt
> > > >
> > > > IPv4CS document is now available at:
> > > >
> > http://www.ietf.org/internet-drafts/draft-madanapalli-16ng-ipv
> > > > 4-over-802-dot-16-ipcs-00.txt
> > > >
> > > > Daniel (Soohong Daniel Park)
> > > > Mobile Convergence Laboratory, SAMSUNG Electronics.
> > > >
> > > > _______________________________________________
> > > > 16NG mailing list
> > > > [e-mail]
> > > > https://www1.ietf.org/mailman/listinfo/16ng
> > > >
> > >
> > > _______________________________________________
> > > 16NG mailing list
> > > [e-mail]
> > > https://www1.ietf.org/mailman/listinfo/16ng
> > >
> >
> > _______________________________________________
> > 16NG mailing list
> > [e-mail]
> > https://www1.ietf.org/mailman/listinfo/16ng
> >
> >
> 

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