Re: [Autoconf] Closing summary on consensus-call for RFC5889modifications

[Alexandru Petrescu <[e-mail]>]

Le 25/08/2010 17:02, Charles E. Perkins a écrit :
> Hello Chris,
>
> Follow-up below...
>
> On 8/25/2010 1:45 AM, Dearlove, Christopher (UK) wrote:
>
>> I didn't make the definition, go back to RFC 2501.
>
> I did. I even quoted 2501 in part of my message to you that you
> didn't answer.
>
>> If the node is to be mobile, attaching at different points,
>> available as a destination to others, it has to be running
>> something.
>
> Mobile nodes do this today and they aren't classified as routers.

Charlie - a Node is a Host "or" a Router - this makes all Mobile Nodes
routers...

I have a hard time seeing Hosts anywhere actually.

I agree that a smartphone with a huge IP stack can do many dynamic
addressing and routing things without running dynamic routing protocols,
w/o running Mobile IP.  And that smartphone is a "Host" in the sense
that it is the end of the application (end-to-end), TCP), end of IPsec,
has GUI to human fingers and eyes, browses the Web as a Client for
entertainment -- all being run sometimes by Routers too... they're all
Routers in the end.

Maybe we could distinguish them better based on the data structures and
algos.

Maybe a Host is a machine whose main goal is to run TCP, whereas a
Router's main goal is to run longest-prefix match.

During one identifiable second a machine doing more MIPS for TCP than
for longest-prefix match, is a Host, otherwise it's a Router.

A machine with more bytes in the routing table than in the TCP
structures, at a given time, is a Host.

Or so...

Alex

  You can even do it without running Mobile IP
> to some extent.
>
> Regards, Charlie P.
>
>
> On 8/25/2010 1:45 AM, Dearlove, Christopher (UK) wrote:
>> I didn't make the definition, go back to RFC 2501.
>>
>> If the node is to be mobile, attaching at different points,
>> available as a destination to others, it has to be running
>> something.
>>
>
> _______________________________________________ Autoconf mailing list
> [e-mail] https://www.ietf.org/mailman/listinfo/autoconf
>