[1st-mile-nm] Subject:Grappa Wireless Internet

[John Brown]

I agree that wireless is a valid 1st mile technology, like any 
technology, when deployed properly.

Things to consider:

Wireless has, at present, limited bandwidth
   Really usable high bandwidth correlates to high $$
   Higher bandwidth tends to shorten the links

Certain flavors require Clear Line Of Sight
Other flavors only need Near Line Of Sight
Environmental issues can be costly
Bad neighbor issues
Requires "science" and proper planing


My personal observations and experience with wireless is that many
folks don't do the proper font side planning.

Owning your own infrastructure, and thus not relying as much on
the local Bell Corp to provide that infrastructure is good.  But
remember that Bell had Bell Labs, which did TONS of thinking
before they deployed something.

If you want to own your infrastructure, you have to be willing to
properly plan/engineer it to reap the rewards.


John Osmon wrote:
> I don't mean to hijack this thread, but I think it is important
> for folks on this list to have a good grounding on what works
> and what doesn't.  Wireless is a legitimate 1st mile technology -- it
> is easier to deploy than fiber and/or copper, but has unique 
> characteristics...
> 
> 
> On Mon, Oct 27, 2008 at 06:49:41PM -0600, Bob Knight wrote:
> [...]
>> I'd be *very* disappointed if I were a paying customer getting the 
>> "performance" you're getting. Sounds like you're paying for a learning 
>> curve.
> 
> In general, WISP customers have to suffer through their provider
> learning the following:
>   - LOS/Fresnel zone issues (how to make the wireless links work)
>   - Interference issues (concentrate on *your* S/N, not what channel 
>     other folks are using)
>   - basic networking issues (route early, route often - briding does
>     not scale)
>   - advanced networking issues (wireless is usually constrained by 
>     packets per second, and *not* bandwidth) 
> 
> It's a lot of things to take in when you're starting out, so a lot
> of WISPs can't climb over the learning curve(s).  (I weep for the 
> customers of some WISPs.)
> 
> 
>>> 5. No system could demonstrate good symmetrical performance average 
>>> was ( performance quoted to actual = Down 60% --- Up 20 - 30% )
>> The very nature of 802.11 makes symmetric performance problematic in 
>> some situations. However, I just pulled 8 megabits upstream from the 
>> WRAP on my roof to one of our servers. YMMV.
> 
> The nature of half-duplex connections make this a sticky subject all
> around.  Capture effects and other issues dictate that you will
> eventually need to move to full duplex as contention issues increase.
> If the WISP has multiple aggregation towers, the backhaul links will
> need to move that direction first.
> 
> 
>>> 6. Security issues were noted on all systems and interception of both 
>>> down and up was fairly easy with open source products
>> Not a surprise. WEP is a joke, WPA isn't much better if one has enough 
>> iv's. That's why I believe in ssh, https and VPN's for stuff I don't 
>> want prying eyes to see.
> 
> End-to-end encryption is *always* the right answer.  Your best friends
> should be ssh/https/VPN/TLS. 
> 
> 
> 
>> We do monitor, and we are putting in place some sophisticated traffic 
>> shaping on a transparent bridge as an anticipatory measure. However, our 
>> DS-3 is being used at no more than 20% capacity down and much less up. 
>> Misbehaving torrents are generally dealt with through education and, 
>> once educated, members tend to be very well-behaved.
> 
> Monitoring is the key.  Tracking and trending some key stats will let
> you *know* when you need to tune/tweak things.
> 
> 
> 
> 
> _______________________________________________
> 1st-mile-nm mailing list
> 1st-mile-nm at mailman.dcn.org
> http://www2.dcn.org/mailman/listinfo/1st-mile-nm
>