[Cu-wireless] wireless anecdotes and the CUW demarcation point

[David Young]

I am at Sascha's house. The wireless network is working very acceptably.
That is, it works very acceptably now that I am sitting on the floor in
the middle of the living room. When I was sitting in the broken chair,
it did not work very well at all.

You thought that I was talking about the C-U Wireless network? No,
I am talking about Sascha's private home network. Works like a charm
when you're in a sweet spot.

By the way, C-U Wireless is working very acceptably, too.

Incidentally, Sascha shares a house with Danielle. Danielle has
been having a hard time getting on-line. Danielle's computer is a G4
Powerbook. The Powerbook is notorious for its short wireless range.
Just like any computer on an 802.11b network, all it takes is few inches'
movement in marginal link conditions, and you are disconnected. No
kidding. Scratch your nose: disconnected. Yawn: disconnected.

Are you beginning to see how devilishly difficult it is to interpret
C-U Wireless problem reports, and to debug C-U Wireless, given only
reports from end-users?

Another thing is that Sascha's AP is not an AP, it is his PC in the
basement. It runs in "ad hoc" or "independent" mode. It seems that
whoever designed the user interface for wireless in Mac OS did not
understand their users or do rigorous user trials. In "independent
BSS"/"ad hoc"/"Computer to Computer" mode, the indicator in the top-right
corner does not distinguish a network of just your Mac (no network at
all!) from a network of two or more computers, and the Mac/Airport will
aggressively "create" a network of just your Mac when there is no other
computer in range. The result? Your Mac gives every appearance of being
"on the net" when it is not.

Consequently, when I was sitting in the broken chair where my Mac's feeble
receiver could barely pick up Sascha's wireless network, the network
indicator at the top of the screen still grinned like I was connected.
Grrr. Lying Mac.

My point with these anecdotes is that C-U Wireless needs to set a rigid
demarcation point (the ethernet jack on the back of the Dell nodes, or
else the jack on the PoE injector on the outdoor nodes), and provide some
standard way for users to verify that on their side of the demarcation
point, everything is working A-OK on the home network. Just off the
top of my head, this might be a web page our node serves on a standard
IP number and/or hostname. Maybe it should contain one of those "DSL
bandwidth measurers."

Thoughts?

Dave

-- 
David Young             OJC Technologies
dyoung at ojctech.com      Urbana, IL * (217) 278-3933