I was just forwarded your thread from a friend who keeps up on the BAWUG
site (my first time here) - I started a Santa Clara based WISP back in 1997
(this was before 802.11b - when 900mhz proprietary cards were the in thing)
and have been running and growing it ever since (using essentially Matt's 
idea of getting 20+ people at $50/mo).  Note however that T1's at $550 are
usually hiding something to get that pricing (like bandwidth metering, limited
IP address range, non guaranteed bandwdith, or some other surprise on the bill)
It's not possible to sell T1 at 550/mo including local loop without loosing 
money - it's either being oversold at the DS3 level, or has some hidden charge.

My base of operations is in Santa Clara near Benton/Kiely (santa clara high
school and Kaiser are the nearby landmarks) - from here I can service a good
portion of santa clara directly - but back in 1997 that wasn't the idea. 
>From the start, I targeted mountain communities who had no high speed solution.
I don't use the web as my primary advertising, but instead send direct mail 
postcards to the addresses that fall within my service area.  Word of mouth
is the most effective advertising out there - neighbors ask each other how
to get internet service, and find out about what I offer, signing up shortly
thereafter.  People living in the mountains around the south bay get to know
each other pretty well and word spreads fast among them.

I started off as hmmwv.net providing service to Redwood Estates (off 17) that
can barely get telephone service (and data at 18-20 kbps due to the line 
problems) - forget cable, dsl, or much else - but santa clara is visable from
most of the south bay mountains, and makes an ideal base of operations 
because the links are all line of site.  A single directional antenna here can
service 200+ customers living on a mountain road who can't get more than just
dialup (or satellite with 3000 ms ping times and < 100 kbps net bandwidth).
Had I tried to service just nearby people, I could not have competed with the
"3mo" or "6mo" specials that cable/dsl sell to attract customers at a low rate
then jack it up after the special is over, or they get sold out.  It was 
important to target the service to a group of people who were certain to
want the service and didn't have any other options except satellite.  
Otherwise, wireless is not cost effective compared to DSL/cable.

When we started the network - we threw in any way to generate revenue to
pay for the T1. Microwave links, colocation (I have a 2000 sq ft datacenter
with 2 ea. 6KW UPS's to run everything) etc.  Later I bought out uwave.com
and added their customer base and a second T1, a 1/2 rack cage in a 4xDS3
serviced colo facility offering 10/100/Gig-E connections, plus a single IP
business DSL with 1.5/384 service and bandwidth guarantee.  (Business rate
DSLs differ from residential in that you have a special support number that
has knowledgable people, service guarantees, and a TOS that specifically
**ALLOWS** the reselling of service, whereas residential DSL and especially
cable companies go out looking for people who are reselling and send them
nasty letters in the mail about violating their TOS, stealing service, etc.)

I've also built a full router and repeater on top of one mountain to allow
me to go 14.5 miles down to south san jose where A/B cable doesn't run and 
DSL is out - it adds 1ms to the hop, but I'm only about 8-10 ms from the major
backbones, so it's still fast going from my core router up to the mountain and
back down to the customer.  One S.J. house has a public block of 32 IPs routed 
to its location where cable/DSL are not available, and a server farm hung off
it.  From that mountain, I have 3 directional antennas aimed into cities with
groups of people not serviced by landline links.  These include parts of San
Jose, Santa Clara, Cupertino, Sunnyvale and parts of Los Altos.

Although everything is 802.11b, I maintain security on the net to not be 
visable unless you are using promiscuous cards and a packet capture program -
the network requires directional antennas carefully pointed to access it and 
doesn't show up in netstumbler. 

What's nice about Matt's concept of having 20 or more people is that once 
you can afford to pay for the monthly recurring charges, you are able to 
gain the advantage of having a critical mass of clients and install 
redundancy (a network that has more than one route out) via NAT.  We 
also have an internal proxy server with a 50G cache disk that catches 
about 80-90+% of the trafic and drops pop-up ads and blinking obnoxious banners
from the data stream.  This means that downloading 20 or 30 MB windows updates
usually takes just a few seconds at 5.5 Mbps (the 802.11b data rate) from 
the proxy instead of 1.544 Mbps from one of the 3 lines into the net.  Also,
should a line fail, the NAT and Proxy server are multi-homed and can select
any 1 of 3 routes out, making the failure invisable to the end user.  Our
firewall also allows assignment of either fixed private addresses for servers
inside the network or DHCP.  With fixed IPs, we can setup dedicated port 
numbers for ssh, etc for people who want to have a server inside the network
that they can get to from work.  We also offer public IP routing via wireless
for people who want that connectivity.  Right now there are 2 class C blocks
available to customers (either colocated or wireless)  This allows servers 
to use 2 IPs on two different routes, then use our DNS servers to implement
failover, improving overall uptime.

We have also expanded into business friendly features like offsite network 
backup to our DLT tape system, a 36" wide inkjet network printer, a highspeed
laserjet postscript network printer, and a network photo printer - even
a networked internet postage meter system for bulk snail mail.  Each are
available on the internal net for just the cost of supplies.  Now we are
experimenting with VOIP systems that offer better quality phone lines to 
remote mountain areas, and phone + data service to homes that do not even
have wired phone service.  Once the wireless infrastructure is installed and
robust, it's amazing what is possible.  One community that shares a common
water tank has a networked water level sensor and a webpage that shows how much 
water they have in the tank.  

If anyone in Santa Clara is interested in trying a link to my Santa Clara 
site directly, email me privately and we can arrange a time to try out a
directional link from my site to yours.  You need a good compas (lensatic 
type with 1 degree marks) and an outdoor directional antenna (the 24 dB
bent grill works well), plus a laptop or equivalent to connect to it up on
your roof while fine tuning it for elevation.  I calculate the magnetic 
heading based on your longitude and latitude either from a map page or GPS.
Most people permanently install a workgroup bridge and an outdoor weatherproof 
box with a voltage regulator to run power over ethernet so that the only cable 
going to the box is a single cat-5, which can be plugged into a house hub. 
This makes a clean, neat install.  With low cost bridges from senao, it's
almost cost competitive with dsl installs, which has long been the problem
selling wireless - the entry cost!

Everett Basham


--
general wireless list, a bawug thing <http://www.bawug.org/>
[un]subscribe: http://lists.bawug.org/mailman/listinfo/wireless

Reply via email to