Re: [DNA-BOF] Draft charter now online

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

Hi JinHyeock,

JinHyeock Choi wrote:
> Dear Nick
> 
> 
>>>This process may occur when link-layer
>>>authentication has completed, a host comes back
>>>into range of a wireless cell, or a cable is plugged in.
>>
>>... or possibly when a device "wakes up" from passively
>>monitoring RAs to actively responding to traffic.
> 
> 
> I think we'd better clarify the notion of 'Network Attachment'. My thoughts 
> are like below. 
> 
> When a node attaches to a network, actually there happen two attachment. 
> First the node makes link layer connection, then it sets up IP layer connection. 

Agreed.
(but see later).


> IMHO, Network Attachment occurs when an interface has made link-layer 
> connection. At this stage, though link-layer connection is estalished, a node 
> may or may not have IP layer connection. 

Since we're just talking terminology here, why don't we talk about
Network Attachment occurring at the network layer (such that
IP datagrams may be sent and received), and Datalink Attachment
as the link-layer operation which allows link-frames to be exchanged.

I'd find it confusing if we were talking about link-layer network
attachment...

> Hence, the node has to detect whether it has IP layer connection or not. And
> if not, it has to gather the necessary information to set up a new IP layer 
> connection to a new IP subent. Which is generally getting a right RA.  

I agree generally that this is the goal, though I get very
wary about providing the wrong impression when people start
talking about 'IP Connection'.

It's a Connectionless medium.

> Hence I think, DNA (Detecting Network Attachment) consists of 
> 1) get a hint of link layer change

This may come because of Datalink Attachment or otherwise...

> 2) check it has IP connection
 >
> 3) gather necessary information for new IP configuration, if its current IP 
>     connection is severed. 
> 
> The above is my basic thought but it needs polishing. Any comments? 

I basically believe that this is what we're talking about with my
previous caveats about IP connections.

I'm not sure that we need to put all that detail (about datalinks)
into the charter though.

It would definitely have a place in the terminology document, and
problem statement(s).

What do people think?

Greg