Re: [DNA-BOF] Detecting Network Attachment BoF description (rev 2)

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

Hi Spencer,

----- Original Message -----
From: Spencer Dawkins <spencer@mcsr-labs.org>
Date: Wednesday, June 18, 2003 10:58 pm
Subject: Re: [DNA-BOF] Detecting Network Attachment  BoF description 
(rev 2)

> Dear Greg,
> 
> I'm assuming this is an INT Area BoF...

I've been working under this assumption too.

> Do we think mobility=tunneling? If so, the impact of network 
> attachment on
> upper layer protocol sessions is substantially reduced, I think?

I see mobility/movement as basically disconnection from one
subnet and connection to another. 
MobileIP/HIP/Tunneling is an important mechanism to reduce 
the impact of movement between IP networks, but I believe that
in the case of host controlled mobility signalling there is an
important role for DNA in achieving movement detection.

This allows tunnel movement procedures to occur. 


> Would this BoF address selecting between interfaces when multiple 
> networkattachments are active, or is this preference assumed to be 
> static?
> Spencer

I think it is important to keep the detection scope to be
per-interface (although it may be useful to tell if two
interfaces attach to the same network).  Most of the address
configuration systems hold per-interface state.

For devices which are moving around, most of the configuration/
mobility management protocols will have definitions of what to 
do with multiple interfaces.  I don't think that DNA would
revisit that.

Greg