Re: draft rolc minutes

[Curtis Villamizar <[e-mail]>]

In message <[e-mail]>om>, [e-mail] writes:
> <<> I agree with the idea of removing IP Encapsulation. The benefits
> <<> of removing IP encapsulation e.g., simplifying the operation,
> <<> protocol independence, better support for MPOA and providing
> <<> a better working ground for the router-router case outweight
> <<> the benefit of retaining IP encapsulation, ie Server Mode.
> <<
> <<I'd like to understand how removing IP Encapsulation provides
> <<"a better working ground for the router-router" case.
> <<
> <<Yakov.
> 
> Removing IP encapsulation virtually implies that all the routers
> in the NBMA cloud be NHRP speakers. Otherwise, the use of NHRP 
> would be limited , in that as soon as your NHRP request hits a non-NHRP
> speaker, it is dropped. Then, in a "homogeneous" NHRP environment,
> your proposals number 1 and number 2, which require that routers
> in the short cut path propagate topology changes information
> can be implemented more easily. As you pointed out, when there are
> non-NHRP speaking routers in the short-cut path, additional measures
> should be taken. I used the term "better working ground" to mean that
> fabric mode provides a more flexible environment towards the implementation
> of some of your proposals. 
> 
> Having said that, among the proposals by you and Bruce, I like the 3rd 
> one the most. Conceptually, it is simple. Also, it is more general in 
> that it will detect a loop regardless of its cause.
> 
> Derya


The problem with the router-router case is routing loops which fabric
mode addesses no better with the removal of IP encapsulation.  It does
allow the routing loops become multiprotocol.

Curtis