[CUWiN] Large Installation

[Nick Inbody]

Jack,


>-----Original Message-----
>From: Jack Unger [mailto:junger at ask-wi.com]
>Sent: Monday, February 28, 2005 11:26 AM
>To: ninbody at co.hood.tx.us
>Subject: Re: [CUWiN] Large Installation


>Nick,

>I'd like to strongly (STRONGLY) suggest that you perform throughput testing
3 or
>4 hops "down the chain" from the point where the network interconnects to
the
>Internet. Do this with multiple users, some along the chain and some at the
end
>of the chain, simultaneously. This is the only way I know to determine
>throughput capabilities (and thus end-user satisfaction) in a large
network.

>jack


My thoughts exactly.  Except build the test bed large enough to push that to
an extreme, maybe even to a breaking point.  Maybe a grid of 3 or 4 radios
wide and 10 to 15 (or more?) radios deep, so there would be some amount of
signal overlap as well as a large number of hops end to end.  Then
throughput test as you described.  It would give an opportunity to create
some worst case scenarios, or at least some bad scenarios in a controlled
environment.  Again, not sure yet if this is feasible in terms of expense.
I have also been looking for a listing of current production PCI cards that
use the Prism chipset, to find one at a low enough price to be "disposable".
Anyone seen a current list?  I am talking to Conexant to see if they can
provide me with one, have not heard back yet.

Nick