Re: router-router NHRP

[Curtis Villamizar <[e-mail]>]

In message <9508030137.AA05015@milliways-le0>, Andrew Smith writes:
> > From [e-mail] Mon Jul 31 14:58:07 1995
> > To: Andrew Smith <[e-mail]>
> > Cc: [e-mail]
> > Subject: Re: router-router NHRP 
> > In-Reply-To: Your message of "Mon, 31 Jul 95 14:21:53 PDT."
> >              <9507312121.AA03245@milliways-le0> 
> > Date: Mon, 31 Jul 95 14:54:31 PDT
> > From: Yakov Rekhter <[e-mail]>
> 
> Yakov,
> 
> > There are two separate issues:
> >    (a) eliminating extra IP hops across NBMA - "short-cut"
> >    (b) determining the Link Layer address at the other end of the "short cu
> t".
> > 
> > What makes these issues distinct is the dynamics of the information
> > that is used. For solving (a) unless the ultimate destination is on the sam
> e
> > NBMA, the information needed to determine a short cut is the routing (IP
> > layer) information. This information is known to be fairly dynamic. For
> > solving (b) we need mapping information between IP and Data Link addresses.
>  The
> > dynamics of this information is determined by how frequent an ATM node
> > moves from one switch to another, or how often the node has to renumber
> > (e.g. due to changing IP providers). This information is expected to be
> > less dynamic than IP routing information.
> 
> That is the opposite of what Norm Finn was saying in an MPOA context a few
> months back: to paraphrase, he asserted that the (b) information needed to 
> respond on a "Bridging" timescale i.e. order of one Spanning-Tree convergence
>  
> time and that this needed to be much quicker than typical "Routing" 
> convergence times for (a). Hence, that routing protocol type of mechanisms 
> were inadequate to deal with (b).

Andrew,

Two points:

  1.  Yakov's b) above is almost compleltely static (the mapping
      between an IP address and its address on the NBMA).

  2.  Some of us couldn't care less what the MPOA is doing, just as we
      couldn't care less what LANE was doing.

> Unfortunately, (a) and (b) are coupled because the correct short-cuts and the
> exit point to use are defined by the Link Layer address given back (the reque
> ster
> knows nothing about shortcuts at all - it merely gets back the exit point's
> LL address without knowing what sorts of shortcuts, if any, went into its 
> calculation). I would hope that the protocol made sure that a requester alway
> s had
> consistent answers to both (a) and (b), even during routing and bridging 
> topology changes: this would be made more difficult if the protocol 
> treated them separately.

The whole point is that NHRP is completely inadequate for a) unless
the tolopogy is constrained such that their is little or no redundant
connectivity to destinations off the NBMA, which some of us consider a
very uninteresting case and very unlikely for a large deployment (what
ROLC originally set out to target).  Somewhere we've lost the
connection between the requirements that this WG set out to meet and
the work output.

> > Yakov.
> 
> Andrew

Curtis