Re: [Roll] Closing on Ticket #10

["Julien Abeille (jabeille)" <jabeille at cisco.com>]

Hi Richard, 

> -----Original Message-----
> From: roll-bounces at ietf.org [mailto:roll-bounces at ietf.org] On 
> Behalf Of Richard Kelsey
> Sent: vendredi 13 novembre 2009 19:14
> To: Michael Richardson
> Cc: roll at ietf.org
> Subject: Re: [Roll] Closing on Ticket #10
> 
> 
> > From: Michael Richardson <mcr at sandelman.ca>
> > Date: Fri, 13 Nov 2009 10:57:33 -0500
> >
> >   So, which requirement does this document satisfy?
> >   Is it: - building routing requirements ...
> >          - home routing requirements   ...
> >          - indus routing requirements  RFC 5673
> >          - urban routing requirements  RFC 5548
> >
> >   I also understand the comment about death by a thousand 
> needle pricks.
> > Will these four stated things be satisfied by a single OFx?  Will 
> > there be four OFx?  Or will many different ones be required for 
> > different situations?
> 
> My hope is that we will need no more than one or two OFxs.
> The reason we need to be flexible now is not that we are 
> likely to need many of them, but that we don't know which 
> ones are the right ones.
> 
> >   If there is no MUST for OF0, then we do not have anything we can
> > interoperate with, and so we do not have an IETF standard.  
>  We might
> > have four standards, and if so, that's fine, then let's be explicit 
> > about this.
> 
> Is there a problem with treating RPL and the metrics document 
> as a single standard for the purposes of interoperation?  RPL 
> can state that there must be a single metric in use without 
> saying which metric it is.
> 
100% agree, RPL+metric+OF should be treated as a single standard. This
is nothing new. What is the point of SNMP if you do not implement MIBs
RFCs? None. A standard does not need to be one RFC. Interoperability was
never met between two device simply because they implement one RFC,
rather that they implement a set of L1/2 standards, L3, L4, L5. 
Julien
> >   We have not discussed security very much yet.
> > 
> >   If there is not space for two OFx (OF0 and OFx) in the code space,
> 
> There is space for two OFx if they are needed.  There is no 
> space for anything that is not needed, big or small.  In the 
> case of something like a consumer thermostat, I do not see 
> any need for having both OF0 and a more nuanced metric.
> 
>                                   -Richard Kelsey 
> _______________________________________________
> Roll mailing list
> Roll at ietf.org
> https://www.ietf.org/mailman/listinfo/roll
>