Re: [DNA] Considerations for DNA Schemes with multiple Interface andLayer 2 Technologies

[Greg Daley <greg.daley@eng.monash.edu.au>]

Hi Pekka,

Pekka Nikander wrote:
>>> Purely DNA point of view, it would be better if CDMA doesn't support 
>>> 'make-before-break' at all. It would make DNA much simpler.
>>
>>
>> I know this was a throw-away line, but it's hard to throw away.
>>
>> I think that MBB will have to be supported, in DNA.  This is especially
>> the case where Unreachability is to be determined from certain tests.
>>
>> We mainly need to be clear about which tests are applicable and
>> what inference is taken from received messages.  I think this
>> is still in coverage for the BCP and Goals though.
> 
> 
> I have to admit that I am totally confused now.  Maybe I just
> don't know the facts.  So, let me ask some stupid questions:
> 
> 1. Are people planning to use make-before-break with CDMA
>    with PPP or not?  Or both?  (If PPP is used, I see no
>    problem wrt. DNA as PPP is then the link layer to IPv6,
>    and the situation has to be handled below the IPv6 layer.)

Indeed, PPP simplifies things.

Most cases we'll be looking at situations with PPP in CDMA.


> 2. Assuming no PPP is used, are there any actual plans of
>    supporting make-before-break between networks that actually
>    have different IPv6 configuration at different base stations?
>    (If not, again, I don't see what we have to worry about.)

 From my understanding, Flarion's wireless technology supports
Make-before break and has a multicast capable MAC.

I believe that these (and other?) technologies are contributing to
802.20.  The Flarion white-papers though, describe a combined 
base-station/router.

This simplifies DNA configuration change testing.

I'm not sure about 802.20, it looks like this is a MAC/PHY group,
so it's conceivable that networks may be bridged to 802.20
in the future.

> 3. Assuming no PPP is used and make-before-break is supported
>    between base stations that potentially have different
>    network configuration, is there any reason why the link
>    layer cannot model the situation as two different interfaces?

At the moment, I can't see why not.

It significantly simplifies neighbour resolution, for the case where
the wired network is bridged behind pairs of base-stations.

> That is, even in the case that the CDMA link layer does not use
> PPP but is directly visible to IPv6, if we can model the radio
> layer connections to different base stations as different interfaces,
> as seen by IPv6, I don't see any problem.  The worst that can there
> happen is that the two interfaces, corresponding to the two radio
> layer connections, are connected to the same IPv6 network.  But that
> is a case that we have to handle otherwise, too.
 >
> There may be some problems related to link layer addresses, etc,
> but someone should really point us to the right documents, or better,
> write a problem statement draft of the situation, trying to clarify
> the current plans of running IPv6 over make-before-break CDMA (and
> other such radio links).  I don't see much use to argue about this
> until we properly understand the details, and at least I can't
> possibly claim understanding them.  I don't even know what documents
> I should read; sorry about my ignorance.

I don't believe there's any ignorance in the matter, but there
are MBB technologies which are multi-access and have different L3
configurations under trial.

The only issue when using different interfaces is that DNA is likely
to be done per-interface, and it may not be possible to get reliable
link-down indications on the 'about to break' interface, even if it's
possible to get link-ups on the 'just made' interface.

Sorry to cause consternation by not thinking through and clarifying
statements.

The problem statement draft is a good idea.

Greg