2447bis and related documents

I've spent some time recently reviewing some related documents:

  draft-ietf-l3vpn-rfc2547bis-01.txt
  draft-ietf-l3vpn-as2547-03.txt
  draft-ietf-l3vpn-as-vr-00.txt
  draft-ietf-l3vpn-applicability-guidelines-00.txt

I have some general questions that may develop into concerns, but for
now may just stem from a lack of understanding and history.

I started with draft-ietf-l3vpn-as2547-03.txt, and must say I didn't
really find the document that illuminating. What I would have expected
from an applicability statement is a "here is where you use this
stuff, and here is where you don't" (perhaps with a "why and why not"
to go along with the statements). Also, given that there are at least
two competing solutions, I'd expect an attempt at making it clear
where one would be used vs. the other, so an operator could easily
understand why there are two approaches and  what the percieved major
differences are.

What I found instead was more of a general overview of the protocol
and mechanisms. A fair amount of the discussion in as2547 is in the
2547bis itself (which doesn't seem like what the goal should be).

Also what, could have been more clear is just what the  assumptions
were for this technology. E.g.:

 From a customer perspective:

  - customer must run an IGP or BGP with SP (in order to inject/receive routes)
  - customer wants SP to handle the routing across backbone, and just
    wants to treat the backbone as an opaque cloud of sorts, where the
    details of how routing is done is handled by the SP.
  - customer doesn't want to deal with setting up tunnels between CEs
    and managing the logicahl interconnects
  - how the SP actually provides the service is something the customer
    doesn't care about (which is I think a true statement in general)

Is there more to the this than the above? Am I missing something
pretty basic?

From an ISP perspective, I wasn't immediately sure why one would
start with BGP, other than perhaps because they already are using it
and understand it well, _and_ that they have MPLS.

Indeed, I'm a bit confused about the way the documents try to say one
doesn't have to run MPLS (citing draft-ietf-mpls-in-ip-or-gre-03.txt)
as if it was NOT MPLS. How can this be? One absolutely must support
the MPLS labeling scheme, or this stuff just doesn't work. That is,
MPLS in IP is still MPLS. Right? (What am I missing here.)

Looking further, I then see that the AS is modeled after: 
      draft-ietf-l3vpn-as-vr-00.txt

which contains a number of questions to answer about a proposed
solutions. Both AS documents do this, but they tend to say hand-wavy
things like "you can run IPsec here" or this or that, as a way to
mitigate.

Have others read the AS and are you happy with what it says?

Thomas

_______________________________________________
Vpn-dir mailing list
Vpn-dir@ietf.org
https://www1.ietf.org/mailman/listinfo/vpn-dir