[MLB-WIRELESS] So how does this routing bit work?

[Ben Anderson]

> look. ethernet frames do not route. neither do 802.11b frames.

route, switch..  whatever language you want to use...  What I'm trying to do
is effectivly do complex 'routing like' decisions at the data-link layer
(where typically switching occurs)...  Much like layer3/layer4 switching.
The worlds fastest 'routers'/'switchs' use this technology.  The frames
could be made to route.  I know they don't now, I'm trying to solve a
problem that exists, that no solution has been built for yet (that I've
found).

> switching doesn't make sense in 802.11b.

Explain?  I think the physical location based 'switching' makes sense.


> layer 2 'routing' (bridging?) does not scale and is link layer dependent.
why
> make life proprietary and difficult by trying to wedge this? oh, btw it
makes
> any solution very not ubiquitous at all.

I can demonstrate why a physical location based 'routing' mechinism will
scale better than one based on passive discovery.
TCP is link layer dependant...  What's your point?
And if it allows a wireless network to scale well beyond what existing
solutions do, why doesn't it make it ubiquitous?


> 802.11b bss mode attempts to help collisions in a broadcast medium, but AP
> handoff protocols are still vendor specific.

I'm talking about replacing all the handoff protocols etc...  Forget bss
mode, etc...  They don't exist within the scope of what I'm talking about.

> IP is ubiquitous and it routes.

I'm not talking about replacing IP.  Never was, never have been.
IP doesn't have a method of discovering routes that will scale to a large
mesh network that's almost completly broadcast based.


> I don't think I can state this more clearly.

Hopefully you have good responses to my response, such that all my concerns
with 'existing' solutions go away, and it'll all just start working.

Ben.

> .vortex
>
> On Friday 29 March 2002 12:52 pm, Ben Anderson wrote:
> > I know what layer 2 switching is...  I was desribing a modification to
this
> > theory, when vortex responded with "scalable role of this IP thingie"
> > Which "IP thingie" is he on about?
> > Why shouldn't I think of "layer 2 'routing'"?
> > And if he's not sure on my position, why is he suggesting I ignore layer
2
> > 'routing' possibilities?
> >
> > Ben.
> >
> > > On Fri, Mar 29, 2002 at 11:25:59AM +0000, vortex wrote:
> > > > not sure what your position is on this, but don't even think of
layer 2
> > > > 'routing'. that's definitely the scalable role of this IP thingie.
> > >
> > > It's generally called 'switching' at layer 2, but the effect is the
same.
> > > An switch uses layer 2 addresses (an Ethernet switch looks at the MAC
> > > addresses), while a router uses layer 3 address (IP addresses, IPX
> > > addresses, etc).
> > >
> > >
> > > Hamish
> > > --
> > > Hamish Moffatt VK3SB <hamish at debian.org> <hamish at cloud.net.au>