Re: Should we share?

[Fred Goldstein <fgoldste@bbn.com>]

Gary K8LT writes,

>RSPF really attempts to address all of the issues that folks have
>mentioned: dynamic routing, subnets split over multiple physical
>segments, mobile users, acquisition of marginal routes, etc.
>
>Some people have been successful in making the current RSPF 1.0
>implementation work, despite known bugs and limitations.  It would be
>interesting to know the network topologies.
>
>Is anyone interested in collaborating to try to implement the RSPF 2.2 spec?

Someone in Australia currently has plans to begin an implementation
quite soon.  We've been discussing some of the technical issues via
mail.  I hope this does come to pass.

RSPF was indeed designed specifically to address the problem of routing
in a loosely-connected packet radio network.  It has specific features
designed for mobile users, routing within subnets, route testing, etc.
As some readers of this group will remember, earlier implementation
efforts (based on the 2.1 spec) were not fully successful.  I attribute
this to the definition of IP "subnet", which implies reliable, full
connectivity below the IP layer.  Since radio isn't like that (unless
you're sharing one big repeater), most AMPRnet areas have NO subnets
at all!  Thus subnet routing is a guess at best; in EMA, it works
only because users are assigned an address based on which router they
see best.

RSPF 2.2 thus explicitly states that there are no subnets; rather,
there are "node groups".  These look the same to most routers, but
when you get down to the local level, they have what we former DECnet
users might call "layer 1 routing" among members of the node group.
(Short horizons handle this.)

AX.25 routing?  Gag me, Witherspoon!  It is like building large
extended LANs.  I sort of make a living now helping clients dig 
themselves out of that hole.
   fred k1io
--------
Fred R. Goldstein   fgoldstein@bbn.com  
Bolt Beranek & Newman Inc.  Cambridge MA USA  +1 617 873 3850