The Routing Over Large Clouds Mailing List Archive by date

[Bruce Cole <bcole@cisco.com>]

> Hi all,
> 
> The nhrp-05 draft specifies the following station behaviour
> 
> "Station S first determines the next hop to station D
> through normal routing processes. If the next hop is reachable 
> through its NBMA interface, S constructs an NHRP request packet
> containing station D's IP address as the target destination"
> 
> Is there any reason why a station could not choose to use
> the next hop IP address as the target destination in the
> NHRP request packet for some IP flows ?

Sure, you're free to attempt to establish a short-cut, or not.

> The spec is written 
> as if stations will always attempt to short-cut - can this 
> be a may instead of a must ?

There is no must in the above text.  A few sentences before the
text you quoted, was:
   An event occurs triggering station
   S to want to resolve the NBMA address of a path to D.

S gets to decide whether or not it wants to do this or not.  The conditions
under which S would want to establish the short-cut are beyond the scope
of the discussion being presented in the quoted section.  One possible
criteria is presented...  I suppose it should be made more clear that this is
the point at which S gets to decide whether or not it wants to even attempt to
establish a short-cut.

> Also let's assume that host station S is aware of its own 
> LIS / subnet. It would thus seem possible to use NHRP for both
> intra-LIS and inter-LIS transfer without the host necessarily
> having a netmask of all 1s, and the consequent undesirable
> proliferation of host routes in the network. Routers would
> be able to do address aggregation, and NHRP could be used
> as a perfectly good intra-LIS address resolution protocol.
> Are there problems with any of this ?

The spec explicitly mentions that address resolution is not limited to
a single LIS.  It also explicitly talks about aggregation of NBMA 
information, including a discussion on possible problems with this (section
6.3).