RE: Draft charter for L2VPN - IPLS & ARP Mediation

At 04:49 PM 5/15/2003 -0400, Hamid Ould-Brahim wrote:

>Ali,
>
> >
> > IPLS has been designed and optimized for VPLS-challenged PE
> > routers. In
> > other words, PE routers that perform IPLS are not expected to
> > perform VPLS
> > functionality.
> >
>
>Not sure I understand your conclusion. IPLS looks designed for
>IP-based CEs. I am not sure why a PE that provides IPLS service
>cannot provide as well VPLS service to other CEs
>(for example LAN switches) - if that PE can handle both -.
>The operator has the choice anyway given the situation and
>what functionality is supported on that PE.

With respect to IP-based CEs, both IPLS and VPLS services look the same and 
they can't tell the difference. However, if end-points are IP, then certain 
optimization wrt to PE internal operation can be achieved.


> > All I want to do is to make sure that operators are O.K. with
> > it. In other
> > words, to have one PE for IPLS functionality and another for VPLS
> > functionality.
>
>If an operator plans to provide IPLS or VPLS
>wouldn't she/he engineer the PE to meet the vpls/ipls requirements?

With VPLS you can offer TLS service to both IP and non-IP end points and 
with IPLS you can offer TLS service to IP end-points only.

-Ali


> >In cases that the operator doesn't need to offer VPLS
> > functionality, then all they need is an IPLS-capable PE. I
> > just want to
> > confirm if that's what the operators expect.
> >
>
>You may be asking fair question, I just don't see why an
>operator needs two distinct PEs if it needs to
>offer both IPLS and VPLS?
>
>Looking at the IPLS draft, I may agree with you that there
>is too much emphasis that an IPLS-capable-PE is a PE
>that cannot provide VPLS service while in fact - in my
>opinion - it should state that if an operator knows apriori
>that the CEs are all IP host/routers, then an IPLS like
>service may be offered (to what extent it simplifies/optimize
>the PE compared to VPLS service needs to be known)...and IPLS
>(as Steven pointed out) is considered just additional tool in
>an operator toolbox (unless in your question you wanted
>to refer/point to something else).
>
>Hamid.
>
>
>
>
>
>
>