RE: draft-morin-l3vpn-mvpn-considerations

[Thomas Morin <thomas.morin at orange-ftgroup.com>]

Hi Dimitri,

Thanks for your comments, a few reply below...

PAPADIMITRIOU Dimitri :
>
> the document should focus on a solid analysis/description before
> concluding with (strong) recommendations that are prescriptive and this,
> at an early stage of the analysis.
> 
> here below a specific example to show that this analytical phase is not
> completed to draw such conclusions/recommendations:
> 
> "By contrast, experience proves that PIM-based mechanisms are
> operationally viable, but with well identified limitations.  Some of
> those limitations may be improved upon (although there is currently no
> clearly defined proposal to do this), however some of the limitations
> appear to be intrinsic to the approach."
> 
> there are stated limitations but which limitations ? and to which extend
> (all protocols developed by the IETF have certain limitations what.
> matters is whether or not they are impacting in the interval of the
> protocol operations and how degradations are impacting outside of this
> interval) 

The limitations of the Full PIM peering solution is something that has
been discussed quite a lot in this working group in the past years. But
you are right in highlighting that we should be more explicit. The aim
is not not repeat in mvpn-considerations many things that have been
already discussed a lot, but I think we can improve the doc by making a
few things more explicit (such as the limitations of PIM, as you
mentioned, but also as pointed out by Eric, on a few other points).

> more interestingly, the statement ack's they may be elevated but since
> these limitations are stated to "appear as" intrinsic to the approach
> and so they can not be elevated ? if this is the case where is the proof
> for such statement ?

We actually say that some limitations seem to offer place for
improvement, whereas some others appear as intrinsic. 

The reference (too implicit, I agree) is to elements such as the
scalability limitations of PIM that may be improved upon by improvements
of PIM such as the refresh reduction worked on by the PIM WG
(non-intrinsic limitation), and such as the fact that the use of PIM for
C-multicast routing doesn't offer the level of consistency between
unicast and multicast that the BGP approach is providing (intrinsic
limitation).

We will clarify this in the document.


> in general also, i suggest that authors keep protocol capabilities
> analysis and consolidate their argumentation from feedback obtained on
> the field as long as BGP-based approaches do not benefit from experience
> on the field. indeed, in its current form, the document reads as
> comparing disadvantages of experimental / deployment experience gained
> from PIM wrt to theoretical expectations/advantages from new BGP-based
> standards.

Not exactly : we do also look at theoretical
aspects/advantages/drawbacks of the PIM approach.  

And as soon as experience is gained from the BGP approach, we should be
able to also talk in terms of experimental experience about this
approach.  

Thanks for your input,

-Thomas



> > -----Original Message-----
> > From: Ron Bonica [mailto:rbonica at juniper.net] 
> > Sent: Monday, December 10, 2007 10:11 PM
> > To: l3vpn at ietf.org
> > Subject: draft-morin-l3vpn-mvpn-considerations
> > 
> > Folks,
> > 
> > Unless I hear an objection before COB Friday, this document 
> > will become
> > a WG draft.
> > 
> >                                       Ron
> > 
> > 
>