RE: comment on draft-takacs-ccamp-revertive-ps-01

["Attila Takacs" <[e-mail]>]

Hi Dimitri,
Thanks for the comments, please see inline!
Attila

> comments on this doc:
> 
> 1. the document should leave outside its scope signaling of 
> the hold-off timer.
> 

So you are suggesting to go with alternative 2.

> reason: this timer is used for multi-layer recovery while it 
> would be advisable to have a more detailed understanding 
> about the needs before extending the protection object
> 

Even in a single layer case one has to set the hold-off timer properly
(e.g., it may be see on a per LSP bases so it can limit the load of the
protection switching process).
I do agree in that muli-layer recovery needs to be properly understood.
On the other hand, from actual extensions point of view, I'm not sure if
there is more to be done than to carry the respective timers. The
complexity of multi-layer considerations relies in the way the proper
values are derived which is probably outside the scope of the extension
doc.

> 2. segment protection: it should address first RFC4872 and 
> then see how it could be applied to segment protection 
> because different segments can have different WTR (i.e. 
> multiple protection object could be included because race 
> condition can appear when protected segment overlap and a 
> failure on another segment occurs on another segment while 
> the other is ready to the reverted).
> 
> example:--A---B---C---D--
>           |   +   |   +
>            -E-+---    +
>               +       +
>               +++++++++
> 
> B-C fails -> recovery segment B-D, then A-B fails 
> 

Agreed, e2e protection seems to be straightforward, while for segment
protection multiple timers may be needed. 

> thanks,      
> -d.
> 
>