The IP Over NBMA (ION) Archive

[Masataka Ohta <mohta@necom830.hpcl.titech.ac.jp>]

Curtis;

> NHRP does offer a solution to the problem of "cut-through" with
> limited ability to scale, much like RIP offers a solution to the
> problem of routing with limited ability to scale.

The problem for NHRP, then, is that ATMARP and intra-subnet
MARS over a LARGE LIS already offer the same degree of the
limited ability to scale with no "cut-through" problem.

Or, do you disagree?

Unlike NHRP, RIP has long history of real-world operation
with no configuration effort required. In local subnet,
I'm still using RIP.

> Rather than offer a counter argument I would like to point out that
> the same argument had been made as far back as the introdcution of
> NHRP to the ROLC WG wrt signalling implosion at a router serving a
> major portion of the non-ATM or non-"the-same-NBMA" portion of the
> Internet.

I don't know the discussion, because I was never interested
in ROLC work to join the WG. But wasn't that a Q.2931 implosion
problem was to be solved by UNI 4.0 LIJ?

Even if not, there could have been optimism of a now-denied
workaround of RSVP merging servers.

						Masataka Ohta