Re: [lisp] LISP Interworking: Proxy Egress Tunnel Routers

["Darrel Lewis (darlewis)" <[e-mail]>]

 > Speaking as an individual, this deployment model has two problems:
> 

To be clear, I also am speaking as an individual.

> 1) The IETF already has other solutinos in that space with a lot more
> successful deployment than LISP (6to4, Teredo, probably even 6rd).
> 
> 2) This deployment model does not actually get to the deployment we
> need.

It is one of many possible ones, and since it is one we have existence
proof for (there are many over the top providers offering v6 via GRE
today), I used it as an example.  I don't think that we're required to
exhaustively discuss the deployment and business models for an
experimental protocol.

But if you like I can provide another example of a virtual over the top
provider offering LISP, VPN, and other services.

> 
> As I understand it, for interworking to be useful, we need to get to a
> point where everyone using LISP either
> 
> A) Is on a link that can loosen URPF
> 

To be specific, the provider has to work around strict uRPF.  The draft
provides a few examples of how to do this, including using a less strict
version, or (preferably) adding a static route to the EID space on the
PE router.

> or
> B)  has a PETR they can use.
> 

Or 

C) they use LISP-NAT

> So, you need to show that there are sufficient incentives to make
> PETRs available to all lisp sites on access networks that are going to
> enforce URPF.  Combining with v6 deployment doesn't really seem to
> help this much.

This is not correct, NAT is always an option.

For v6, we have a difference of opinion here it seems.  I believe that
using this to hop over my provider's v4 only access network is
interesting.  I don't see why the existence of alternatives should
preclude LISP experimenting with this problem - certainly there are
other alternatives to the routing problem (like shim6) that have not
precluded experimentation with LISP.

> 
> I can see incentives for whomever is selling you your EID to make a
> PETR available to you, but unless you are going to end up tied to a
> single ISP, I don't see why they will be in a position to make a PETR
> available close to you.

If we can see the incentives, that should be enough correct? 


-Darrel