[TriLUG] OT: solar wireless mesh router

[Magnus]

Tim Jowers wrote:
> Anyone know the story of the Town of Cary fiber? Way I
> heard it they never lit it up.

Link?  I hadn't heard about this.

> 4) No port filtering/blocking

I honestly think this point should be a premium service.  People who
don't need it shouldn't have it.  I'm thinking mostly of Windows zombie
machines, spam relays, etc.

I don't really have a problem with forcing outbound SMTP through a
well-maintained relay and blocking residential customers from sending
directly from their own home smtp servers.

And with the basic service it would be smart to block a lot of services
just to protect the ignorant masses.  Pay an extra $5 a month and have
the filters removed and get a static IP.

> Neither DSL nor cable comes close to the performance of a T1.

They often exceed the bandwidth but rarely do better on latency.  It's
the low latency of the T1 that is likely impressing you.

> $400/month T1 plus equipment amortization.
> 20 users at $20/month. 1/2 of the price of TWC. Probably as good or better
> actual performance.
> Can get a T1 with a 1 year contract.

You're going to need to subscribe more customers than that at those rates.

And it's still just a T1.  If nobody else is doing anything, the best
you're going to get is 1.54Mbps.  Downloading the latest Fedora ISO's is
going to take a looooong time.  As bad as TWC is, I can still download
most ISO's pretty quickly.

The T1 is a dead end, really.

> Some enticement may be necessary to get people to add 100ft towers in their
> yards, or rent space on buildings (cost prohibitive for a low caost network
> IME).

I spent quite a lot of time when I was in Philadelphia trying to
kickstart a grassroots wireless mesh network.  I had secured permission
to mount routing nodes on top of some of the tallest buildings for free,
as well as taller buildings in the surrounding counties that had a clear
LOS to the Philadelphia skyline.

The practical limitations of a wireless mesh killed the idea in the
crib.  The stark reality of the situation is you really need to run wire
or fiber to make a go of it on any scale beyond half a dozen or so
friends who live within LOS of one another.

> What is the cost of business T1 or business cable today? 

T1's cost at a minimum about $300 a month, but more likely closer to
$500 around here.

Business cable is a little over $100 a month last time I checked.

> Could be the
> business customers underwrite the entire network cost.

You're going to have a chicken & egg problem. Nobody is going to want to
get on the network until it's up and running, peered to a number of
ISP's offering an excellent value.  And that's not going to happen until
you've got a number of people on it.

The tack I was taking with my own mesh concept was to make it a hobbyist
network AT FIRST.  Make it experimental at first for pure geekery sake.
 Then stabilize the service, and start offering some real advantages to
it over the cable modem paradigm.  Then start enticing our less-geeky
friends & family to participate in the mesh.  Then when you hit critical
mass, start approaching the local data centers, universities, etc for
peering arrangements, etc, etc.  The businesses don't really get on
until there is a compelling reason for them to do so.  Long-term
speculation of what MIGHT come of the service is not compelling, and as
there is very little precedent for something like this being successful
over the long haul, it's not a good business investment for most.

> Phase 3: Established a MAN. 

Honestly this should be phase 1.  It doesn't have to be big.  But the
planning should be laid out early on to provide for city-wide scalability.