Re: FW: draft-zhang-mpls-interas-te-req-01.txt

[raymond zhang <zhangr@info.net>]

Hi Bert,
Thanks for passing the comments.  The inter-AS TE issues we've been 
wrestling with (both cases) for the past years drove our thoughts toward 
the inter-AS TE as documented in this draft today.

Hi Kurtis,
I am very glad you echoed to our thoughts that it is the right time now in 
solidifying inter-AS TE requirements.  We have discussed  Inter-AS 
requirements in various forms with quite a few folks from SPs in Europe, 
North America and Asia Pacific for over a couple of years now and a set of 
general issues listed in the following:

1. How to guarantee bandwidth along a QoS path consisting of multi-AS 
segments (including SLA boundary issues) ?
2. How to achieve reliability across inter-AS links/nodes (FRR requirements)
3. How to support and manage TE across different SP admin domains ?
4. Inter-AS resource optimization internally to a multi-AS SP's network.
5. How to structure business inte-connect agreements such that it would 
make sense for both SPs with interconnect requirements.

#5 is not the issue for TEWG to look at but #1, 2, 4 and part of #3 
certainly are...

See more comments in line please below:


At 08:17 PM 3/10/2003 +0100, Wijnen, Bert (Bert) wrote:
> More w.r.t. the specific question on if this is
> timely and usefull work.
>
> Thanks,
> Bert
>
> -----Original Message-----
> From: Kurt Erik Lindqvist [mailto:kurtis@kurtis.pp.se]
> Sent: maandag 10 maart 2003 17:06
> To: Wijnen, Bert (Bert)
> Cc: Randy Bush; ops directorate
> Subject: Re: draft-zhang-mpls-interas-te-req-01.txt
>
> ...
>
> > Kurtis, thanks for the review.
> > Now... besides this specific document (which is a possible base
> > to start work in TE-WG), the specific question I would like
> > to see answered by you guys from the real OPERATIONAL side is:
> >
> >    Is it timely to work on inter-AS TE requirements now.
>
> I would say yes. But I would also like to see some work done on
> non-MPLS solutions.
>
> > It seems there are even two forms of it:
> >
> >    Inter-AS but within one (global) provider.
> >
> >    Inter-AS between multiple providers
> >       (maybe that is better termed Inter-Provider)
> >
> > Comments?
> >
>
> Both exists, and I worked with both problems. I don't think that they
> are different enough from a technical perspective to make to cases.
> However, from a implementation policy POV they are very different where
> the second is the hard one.
>
> The problem that need to be looked at is as well is if this create more
> complexity than what you gain from implementing it.
Good point...  In section 6.2, we talked about the scalability which 
implicitly addresses the complexity issue for any proposed solutions.
We will expand this issue a bit further in the next revision.

> I think that the
> VPN model given in the draft is a good model, but it is at the same
> time way to simple, as I tried to show. Also, I think that starting
> with a non MPLS network is easier, and would give hits at what needs to
> be done for MPLS TE.
The initial objectives of this draft is to be able to address these issues 
within our deployment time frames.  Therefore we took on the initial 
approach of extending current MPLS intra-AS TE to Inter-AS since there is 
already quite a bit of deployment base and operations experiences  with 
intra-AS MPLS TE.  I am not sure if there is any known IP-based TE 
approaches addressing the above issues today and we may certainly consider 
the inter-as requirements for non-mpls or IP-only based networks in a later 
stage as we learn more from those SPs with non-mpls inter-AS TE requirements.

Thanks,
Raymond
> - kurtis -