Re: Comments on NHRP spec.,, modes of deployment etc.

[Bruce Cole <[e-mail]>]

> NHRP spec authors:
> 
> I am confused by the way the current version of the spec. is worded 
> in that it appears that modes of deployment are described as different ways
> in which servers may interact.

The current plan is to remove server mode from the spec, as server mode (as it
is currently described) falls apart when the IP network layer encapsulation
is replaced by an LLC/SNAP encapsulation.  Removal of server mode should
eliminate the confusion, no?

Note that there are several side effects of this encapsulation change
which have not been discussed/emphasized on the list.  Some which come to mind
are:

- No more requirement for a router-alert option, or the equivalent.
  Each hop within the NBMA now either supports NHRP, or (should!) fail to route
  the NHRP traffic.

- Partial deployment of NHRP more problematic
  An intermediate hop which does not support NHRP causes NHRP address
  resolution to now fail.  It should still be possible for a subset of the
  cloud to make use of NHRP...

> I would like the spec to define and differentiate between the NHRP
> client-server interface

I see no difference from a protocol point of view.  Any station within
the NBMA can potentially send both request and response packets.

> and the NHRP server-server interface, and keep
> these separate from any discussion of modes of deployment.

There is no server to server protocol.