Re: Polling for new WG I-Ds

[Richard Rabbat <richard@us.fujitsu.com>]

Ben,
see inline.

benjamin.niven-jenkins@bt.com wrote:
> Igor,
>
> owner-ccamp@ops.ietf.org <> wrote:
>   
> > Deborah.
> > (My hat off since I am talking to a lady)
> >
> > Have you been in a situation when you just performed a
> > software upgrade and a network operator tells you that your
> > new release sucks and he wants the network in the state of
> > this morning prior to the upgrade? No? Than you are lucky.
> >     
>
> I'm not sure I understand the relevance of this.
>
>   
> > Besides, what is the problem with moving "hundreds/thousands
> > of LSPs from CP->MP"?
> >     
>
> The operating tenet is "if it aint broke don't fix it" so it'd likely
> have to be a very good reason in order to make an operator move
> hundreds/thousands of LSPs from CP->MP or MP->CP.  
I guess the jury is still out w/r to CP->MP as the discussion on the ML 
shows; but w/r to MP->CP, when an operator decides to deploy a CP, then 
control functions that resided in the MP should be delegated to that CP. 
There needs to be a migration strategy for that operation.

> It'd also have to be
> something forseeable as I can't see operators moving their LSPs from
> CP->MP routinely when doing a software change.
>   
I believe the authors were concerned about error handling and are not 
talking about this as a routine operation

> Ben
Richard.