Re: router-router NHRP

[Dimitry Haskin <[e-mail]>]

Yakov,

> Subject: Re: router-router NHRP 
> Date: Wed, 08 Nov 95 18:39:00 PST
> From: Yakov Rekhter <[e-mail]>
> 
> Dimitry,
> 
> > I don't believe you've answered my question on the *implication* (not
> > intention) of the "first NHRP target constraint".  To reiterate,  does
> > this constraint require that a Request can be only originated if
> > a route to the requested target is already present in the originating
> > forwarder FIB?  And if so, wouldn't it mandate a large FIB in some cases?
> 
> This constraint does not imply that the originator of a Request has to
> maintain in its FIB a route that covers precisely the same set of
> destinations as the target, priot to issuing the Request.
> For example, the originator of a Request may have in its FIB a route
> to 192.9.200/24, but may generate NHRP Request for 192.9.9.200.9
> as the target.  
> 

It seems that we fail to communicate.  I never said that your "first
constraint" requires the originator to "maintain in its FIB a route
that covers *precisely* the same set of destinations as the target".
I said (and I believe that you've just confirmed it) that it requires
the originator to carry *a* route to the target which, of cause, can
be a supperset of the target.  If we take as an example a transient
domain with a need to carry the complete Internet set of target, it can
account for circa 30,000 routes to cover the reachable destinations.

Dimitry