Re: [RAM] Comment on draft-farinacci-lisp-00.txt (LISP)

[marcelo bagnulo braun <marcelo at it.uc3m.es>]

sorry for the delayed reply...


El 01/04/2007, a las 1:18, Tony Li escribió:

> > >
>
> > > I'm not yet convinced that this is really a requirement.  Consider  
> > > what you're asking: the ability to TE on a per-site basis, given 
> > > that  traffic is already assigned to PA prefixes.  Typically in the 
> > > past,  ISPs have wanted to do TE within another providers prefix, 
> > > and thus  we have more specifics.  This is incredibly important for 
> > > managing  peering ratios and the like.  However, applying TE on 
> > > multi-homed PI  sites seems like it would be much less important 
> > > requirement.
> >
> > but this seems to be the only difference from the operator 
> > perspective, between a PI-BGP based multihoming solution and a 
> > multihoming solution based on PA and an host based mechanism like HIP 
> > or shim.
>
>
> Not necessarily.  Consider an ITR/ETR co-located with a PE router.  
> That's very interesting from an operator perspective as they can then 
> effectively change locators for a site without site intervention.

i am not sure how this would work...

are you considering the case of a multihomed site connected to two or 
more different ISPs?
because if this is the case, it seems a bit strange to me that a end 
site will allow one of its isps to select the other ISP to send traffic 
(by selecting the other ISP prefix/locator)

i mean, i think that allowing a ISP to select the other ISP locator may 
not be such a good idea and seems a bit off limits of what a ISP should 
be able to control on its client's equipment...

am i missing something?

> > since the ISPs claim that there is some TE functionalities missing, 
> > it zould be important to understand what are those so ze can actually 
> > provide those...
>
>
> I agree.   Moreover, it's important that we confirm that the features 
> that we are providing are those that are beneficial.

agree

but it seems that so far, we haven't managed to get the operational 
community to describe what exact TE capabilities they need (except from 
J. Schiller initial effort a year ago, but that didn't end up 
documented in a draft AFAICT...)


regards, marcelo


> Tony


_______________________________________________
RAM mailing list
RAM at iab.org
https://www1.ietf.org/mailman/listinfo/ram