Re: [core] sleepy nodes

Hi, perter,

"Taking the advice of Zach, that you only want to define a proxy leaves
also many possibilities open, like: ...... "

Could you give further explanations about the proxy.
Is the proxy a resource constrained node?

Regards,

Gengyu

-----原始邮件----- 
From: peter van der Stok
Sent: Tuesday, October 15, 2013 7:49 PM
To: Core ; akbar rahman
Subject: [core] sleepy nodes

Dear all,

reading the draft on the need for sleepy nodes by Akbar, I become more
convinced that the subject is not right for IETF.
Already defining a sleepy node is fraught with application oriented
aspects.
Taking the advice of Zach, that you only want to define a proxy leaves
also many possibilities open, like:

Does the sleepy node communicate only with the proxy?
Can the sleepy node send to other nodes than the proxy via multicast or
unicast?
Does the sleepy node need maintenance?
Is the sleepy node initialized at the factory and it never listens to
any messages afterwards?
Or is there an initialization capability, eventually supported by a
battery recharge.

I can go on and on like this, (for example on timing constraints) and
none of the questions will get satisfactory unanimous answers.
Therefore, I remain with my conclusion that sleepy nodes are interesting
but too much conditioned by application constraints to define a proxy
approach that is generally applicable.

If anything needs to be done I still think that the "mirror server"
draft is the best choice because I recognize the boundary conditions.
But the latter is a very application biased opinion.

Peter

-- 
Peter van der Stok
mailto: consultancy@vanderstok.org
www: www.vanderstok.org
tel NL: +31(0)492474673     F: +33(0)966015248
_______________________________________________
core mailing list
core@ietf.org
https://www.ietf.org/mailman/listinfo/core 