The IP Over NBMA (ION) Archive

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

> [Please ignore, if this point has already been discussed before; in
>  that case I'm sorry for wasting the bandwidth & time.]

Don't mind. It's a lot better than blindly believing in NHRP and
waste bandwidth & time in purposeless standardization effort.

> ...NHRP scalability...

No, NHRP inscalability. :-)

> but the point is can NHRP protocol send reply on load-computation basis.

Do you think that system performance should degrade at heavy load
when the performance is needed most?

> So that if the short-cut can't be made to the destination directly, at
> least it could be made to *some-point(one/two/few hops short of 
> destination)* along that path.

It seems to me that such an incomplete path will be a cause of
persistent loops.

Moreover, with highly logical LIS configuration where shortcutting
could be meaningful, one hop shorter path can be very inefficient.

For example, if the last hop LIS is configured around the world,
your packet directed to the next room host may first goes to the
last hop router on the other side on the earth.

							Masataka Ohta