Other SDOs who have control over their full architecture and its plumbing
can define this on their own. 

We can talk about it in the DSL-specific I-D. But the level of details in
that informational document is not likely to match what DSL Forum may
eventually define in their own standard specification. 

Alper

> -----Original Message-----
> From: Mark Townsley [mailto:[EMAIL PROTECTED]
> Sent: Tuesday, October 10, 2006 12:35 AM
> To: Alper Yegin
> Cc: 'Yoshihiro Ohba'; [email protected]
> Subject: Re: Network Layer (was RE: [Pana] Other suggestions for pana-
> pana)
> 
> Alper Yegin wrote:
> > The PAA will know the IP address of the PaC, but that's only for
> reaching
> > the PaC for PANA signaling.
> >
> > Like Mohan was saying, we'd leave filter details out-of scope for the
> base
> > specification.
> >
> Where would you define this, for example, for the DSL case? In a
> separate DSL-specific draft?
> 
> - Mark
> > Alper
> >
> >
> >
> >
> >> -----Original Message-----
> >> From: Mark Townsley [mailto:[EMAIL PROTECTED]
> >> Sent: Monday, October 09, 2006 11:28 AM
> >> To: Alper Yegin
> >> Cc: 'Yoshihiro Ohba'; [email protected]
> >> Subject: Re: Network Layer (was RE: [Pana] Other suggestions for pana-
> >> pana)
> >>
> >>
> >> OK, are you still planning on being able to identify at least one PaC
> IP
> >> address for the EP to setup a simple filter based upon? Will this just
> >> be the IP address that PANA packets arrive at the PAA from?
> >>
> >> - Mark
> >>
> >> Alper Yegin wrote:
> >>
> >>> Mark,
> >>>
> >>> Your earlier recommendation to deal with only IP addresses (scrapping
> >>>
> >> MAC
> >>
> >>> addresses) as the device ID would cause complications. As Yoshi
> >>>
> >> explained
> >>
> >>> earlier, dealing with router type PaCs, PaCs with multiple addresses,
> >>>
> >> PaCs
> >>
> >>> with changing addresses (for privacy), etc. becomes a burden.
> >>>
> >>> I think one way to deal with your concern is to remove the "device ID"
> >>>
> >> from
> >>
> >>> the PANA specification. That way we don't need to be concerned whether
> >>>
> >> it is
> >>
> >>> an IP address, a MAC address, or a locally-significant identifier.
> >>>
> >>> Device ID is used for:
> >>>
> >>> - PaC's EP discovery.
> >>> - Per-PaC access control filter settings on the EP.
> >>>
> >>> I think both can be handled by mechanisms outside PANA specification.
> >>>
> >>> With that, we can also remove the PaC-EP- master key computation.
> >>>
> >>> We still have requirements about "device IDs" in RFC 4058. But that
> >>>
> >> should
> >>
> >>> not be a concern if we want to make this simplification.
> >>>
> >>> Alper
> >>>
> >>>
> >>>
> >>>
> >>>> -----Original Message-----
> >>>> From: Mark Townsley [mailto:[EMAIL PROTECTED]
> >>>> Sent: Friday, October 06, 2006 5:27 PM
> >>>> To: Alper Yegin
> >>>> Cc: 'Yoshihiro Ohba'; [email protected]
> >>>> Subject: Re: Network Layer (was RE: [Pana] Other suggestions for
> pana-
> >>>> pana)
> >>>>
> >>>> Alper Yegin wrote:
> >>>>
> >>>>
> >>>>>>> The above two issues are related. If PANA would eliminate trying
> to
> >>>>>>> operate on L2 access controls, and rely solely on an IP Address
> and
> >>>>>>> associated filter, then it would truly be a Network Layer
> >>>>>>>
> >>>>>>>
> >>>> authentication
> >>>>
> >>>>
> >>>>>>> and Network Layer access protocol. I believe this would go a very
> >>>>>>>
> >> long
> >>
> >>>>>>> way to eliminate confusion over where PANA can add value. I cannot
> >>>>>>> understate this enough.
> >>>>>>>
> >>>>>>>
> >>>>>>>
> >>>>>> This is fundamental difference in how we view PANA.  I don't
> consider
> >>>>>> PANA as "Network Layer access protocol" while I only consider PANA
> as
> >>>>>> a network layer protocol for carrying authentication information
> for
> >>>>>> network access.
> >>>>>>
> >>>>>>
> >>>>>>
> >>>>> This is a terminology issue, I believe.
> >>>>>
> >>>>> As we said in RFC 4058:
> >>>>>
> >>>>>    After a device is authenticated by using PANA, it MUST be
> >>>>>
> >> authorized
> >>
> >>>>>    for "network access." That is, the core requirement of PANA is to
> >>>>>    verify the authorization of a PaC so that PaC's device may send
> and
> >>>>>    receive any IP packets.
> >>>>>
> >>>>> So, PANA authenticates for (and used for authorization of) "network
> >>>>>
> >>>>>
> >>>> access."
> >>>>
> >>>> "Network" access to me means access to send "Network" packets, which
> is
> >>>> synonymous  to "IP packets" in the statement above. Expanded a bit, I
> >>>> believe the intent here is to say that:
> >>>>
> >>>> PANA authenticates a PaC so that it can be authorized by a PAA to
> >>>> install filters on an (colocated or separate) EP that permits IP
> >>>>
> >> packets
> >>
> >>>> to flow to or from a PaC on a given network. I see nothing in that
> >>>> statement that requires L2 or L1 specifics. Since the EP probably
> >>>> already has to filter at the network layer simply to permit/deny the
> >>>> PANA packets themselves onto the network, it is perfectly reasonable
> to
> >>>> assume that enforcement may be done at L3. Doing so, reduces the
> >>>> complexity of your problem dramatically, and I believe puts PANA in a
> >>>> place that most people will be able to readily understand.
> >>>>
> >>>>
> >>>>> Execution of access control is orthogonal to PANA authentication. A
> >>>>>
> >>>>>
> >>>> network
> >>>>
> >>>>
> >>>>> can choose to change IP filters to allow traffic from authenticated
> >>>>>
> >>>>>
> >>>> client,
> >>>>
> >>>>
> >>>>> change L2 filters to do equivalent, or even do some L1 filtering.
> >>>>>
> >>>>>
> >>>>>
> >>>> Which is where you get into the layer violations that people get
> >>>> concerned about, and why your protocol ends up looking more complex
> >>>>
> >> than
> >>
> >>>> it need be. Stick to "Network Access" and you are fine. Move down
> into
> >>>> "Data Link Access" and you start putting the cart before the horse,
> and
> >>>> open yourself up to an array of complexities by looking downward into
> >>>> the stack.
> >>>>
> >>>>
> >>>>> At the end, what's really allowed/disallowed is the IP network
> access.
> >>>>>
> >>>>>
> >>>>>
> >>>> If you allow this type of logic, I could say that in the end what is
> >>>> really allowed/disallowed is email access. If you are carrying PANA
> >>>> packets on the network, by definition you already have at least some
> >>>> level of data-link and Network (IP) connectivity. I honestly believe
> >>>> that trying to manipulate L2 and L1 via PANA is becoming a burden
> that
> >>>> may, in the end, sink the entire ship.
> >>>>
> >>>> - Mark
> >>>>
> >>>>
> >>>>> I hope this clarifies.
> >>>>>
> >>>>> Alper
> >>>>>
> >>>>>
> >>>>>
> >>>>>
> >>>>>
> >>>>>
> >>>>>
> >>>>>
> >>>>>
> >>>>>
> >>>>>
> >>>
> >
> >


_______________________________________________
Pana mailing list
[email protected]
https://www1.ietf.org/mailman/listinfo/pana

Reply via email to