Re: Draft charter for L3VPN: inter-AS/inter-provider

Alex,

Comments inline:

On Fri, 9 May 2003, Alex Zinin wrote:

> Folks-
> 
>  Regarding inter-AS and inter-provide. Small clarification: the
>  proposal is not to move it outside the charter completely, but
>  to clarify what the WG would be doing first and what next, i.e.,
>  make visible progress on basic things and then move to more complex
>  scenarios.
> 
>  Now, having read postings from everyone, it seems that the agreement
>  is (correct me if I'm mistaken) that inter-provider is better
>  put as a future item for now.
> 
>  Regarding inter-AS single provider, I have a couple of questions:
> 
>  1. Jim Guichard said:
> 
>       > 2. Is complexity associated with inter-AS (1 provider) scenarios
>       >    sufficiently higher than for the single-AS case?
> 
>       potentially yes for the same reasons as (1). Jim
> 
>     Two reasons have been brought in (1):
>     
>       a) different encaps for 2547 (MPLS/GRE)

There is no fundamental difference with regards to supporting different
encaps in the single AS and inter-AS case. 

>       b) provisioning end-to-end LSP is more complex in the multi-AS
>          case
> 
>     It seems that this may indeed complicate the mechanisms.
>     I'd love to see a little more discussion on this.

2547 describes the mechansims for inter-AS. IMHO they are well understood.
Set up of a RSVP-TE inter-AS LSP is work in progress. However
2547 does not need any additional mechanisms for using a RSVP-TE inter-AS
LSP between the end PEs.

> 
>  2. Are we going to need additional documents compared to what we
>     already have in order to describe inter-AS scenarios
>     appropriately? 

Base mechanisms for implementing and deploying inter-AS are in place and
are described in 2547. 

If yes, I'd rather say one-AS first, then multi-AS
>     single provider, and then multi-provider.
>

I would still say one-AS and multi-AS to begin with.

Regards,
rahul
 
>  Regards.
> 
> Alex
> 
> 
> 
>