On Mon, 30 Jul 2012 00:21:03 +0000 (UTC), danny burstein
>We've got a cul-de-sac with a half dozen homes, and
>are interested in getting a high speed connection
>and sharing it. Our local cable provider (Charter)
>provides 30 meg at a price that would work out.
We have both Charter and Comcast in the area. Charter sucks. Comcast
>(most of the folk here are part timers so the speed
>should be adequate when shared)
Famous last assumptions. I used to use this rule of thumb for number
of clients per access point.
100 home users doing email and web surfing
10 business users doing whatever business users like to do.
1 file sharing user
Divide by 2 for watching TV over the internet (Netflix etc).
However, that's not your real problem. It's going to be policing the
system to make sure it's usable. It only takes one user with a virus
to saturate your OUTGOING bandwidth. You'll need to monitor the
traffic, monitor the type of traffic, and be able to do something
about abusers, freeloaders, hackers, and over-users. Creating the
ultimate fair cost sharing model is also a challeng. Some users are
just going to use more than others.
However, all of that is minor compared to getting a phone call from
one of the users at 2AM in the morning asking "Is the Internet up"? If
you set yourself up as the keeper of the system, that's almost
guaranteed. Even if you charge for service calls, everyone assumes
that a 2AM phone call is free.
How to start a WISP:
Ubiquiti WISP guide:
Mikrotik WISP guides:
>Placing a Linksys model mumble mumble five y/o in the
>windowsill almost gets enough connectivity, so we figure
>that putting it up on the roof would work out.
Maybe. Do a site survey and see how many other 2.4Ghz systems you're
going to hear. Higher isn't always better as you just end up hearing
more interference. If you can handle the cost, this is a good time to
seriously consider using 5.8GHz (802.11a) using stand alone wireless
clients with ethernet connections. If the users need more
connections, they can use their own home 2.4GHz wireless routers.
Ubiquiti makes (usually) good hardware.
Try Nanostation 5 or Bullet 5 for the client radio. The bigger and
uglier the antenna, the better it works. Mikrotik also makes good
There are a tangle of blogs, forums, Yahoo groups, etc for these
companies. Dive in.
Your window sill link test isn't good enough. You have to be able to
move a few megabits/sec with a low error rate in order to declare it
working. That's not easy, especially if there's co-channel
>What we're looking for is a unit that's reliable and
>can handle the multiple connections - possibly up to 50...
With 50 full time connections, you'll probably need to divide the
traffic between multiple access points. Two or three Bullet 5 series
radios, with 3 seperate (sector??) antennas, should work.
>Bonus points for something that's powered by POE, so we
>can run low voltage up there.
All the Ubiquiti radios are PoE. However, they're NOT 802.11af
compliant, so you'll need to buy the proprietary Ubiquiti PoE adapter.
>Prefarbly something intrinsically designed for outdoor use
>with a professional enclosure. Alternatively we can
>build a weatherproof box...
A weatherproof box might prove useful if you're going to do solar
power and put your batteries and router in the box. If Charter will
only give you a single IP address (highly likely), you'll need a
router between the Charter cable modem and access point radios. If
anything is likely to fail, hang, or screwup, it will be this router.
I like Sonicwall, but you probably will not like the price tags. They
also have expensive user licenses, which adds more cost. Something by
Cisco should also work, but I don't know much about their current
>We realize this won't be fifty dollars...
I don't think you realize how expensive this can become. It's not the
hardware. It's the support hassle that costs time and money. If you
have 50 users, and Charter goes down in the middle of the big football
game, you'll get 50 phone calls. If it's raining, you'll be expected
to fix things immediately. Do you REALLY want to do that?
Hint: Do the math. If it doesn't work on paper, it's not going to
work when you build it.
Jeff Liebermann email@example.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558