Wifi Highway - could this work?

James Rogers jamesr@best.com
03 Jan 2003 12:18:49 -0800


On Fri, 2003-01-03 at 09:27, Eugen Leitl wrote:
> 
> Out of this only email or similiar high-latency
> messaging is not latency bound, provided the timeout is jacked up high and
> the app can deal with IP address mutating under its feet (most current
> apps can't).


A good approach is to leave the IP alone and let Layer-2 worry about
delivering the packets to a moving target.  Layer-3 doesn't need to know
that you are running a wild-n-wooly network at Layer-2.  Most of this
kind of technology is really young, but it is being worked on and some
usable first draft implementations exist.  In a couple years good
implementations should be ubiquitous, and the nice thing is that it is
backward compatible with existing Layer-3 apps.

It doesn't solve every problem, but there is a lot of latitude in the
kinds of games you can play at Layer-1 and Layer-2 while keeping the
gymnastics transparent to Layer-3.  There are hardware vendors
interested in actively supporting and developing these types of
capabilities but they need some direction from the people who are
actually deploying networks that need this type of capability.  Most of
the older network providers can't use this technology (too much
dependence on Layer-3 manipulation for basic tasks), so it is up to
newer companies to drive it.


-James Rogers
 jamesr@best.com