Re: [DNA-BOF] Attachment Detection charter text for discussion

[Brett Pentland <brett.pentland@monash.edu>]

Hello.

montavont@clarinet.u-strasbg.fr wrote:
> Hi,
> 
> Quoting Brett Pentland <brett.pentland@monash.edu>:
> 
> 
>>Jari Arkko wrote:
>>
>>>Brett Pentland wrote:
>>>
>>>
>>>>Greg Daley wrote:
>>>>
>>>>
>>>>>"Network Attachment occurs when a host arrives on
>>>>>a new IP subnet. When attaching to a network, a
>>>>
>>>>
>>>>
>>>>I'm not sure that "when a host arrives on a new subnet" is the
>>>>correct way to define this.  Perhaps something along the lines of
>>>>"when an interface becomes active on a link instance" with an
>>>>appropriate definition of "link instance" in the terminology
>>>>document.
>>>
>>>
>>>Does this formulation require that the link layer
>>>always be able to deliver up/down information to the
>>>IP stack? I realize that for fast movements this is
>>>probably a requirement anyway, but there may be
>>>other cases where lack of hints could be tolerated.
>>>
>>>--Jari
>>>
>>
>>Perhaps my wording is not so good after all.  I simply meant that
>>"network attachment" occurs when an interface becomes active on a
>>link instance.  How that is detected is the work of DNA.  It may be
>>hinted at by an L2 trigger, or perhaps is not noticed for a little
>>while until an RA is received.  Making the most of the available
>>information and determining exactly what it means (with respect to
>>the need for configuration changes) is what I think we need to work
>>on.
> 
> 
> Totally agree. I find the wording very well ("when an interface becomes active
> on a link instance") but maybe we could add that the way the link is discovered
> is the work of DNA and can be done in many ways from a L2 trigger to the
> reception of a RA.
> 

The additions sound fine to me.  There has, however, been some
discussion on the list about the definition of "active".  Taking
some words from a side discussion with Greg, perhaps we could
say something along the lines of "when an interface becomes able
to send and receive IP datagrams on a link instance".

Brett.