"rob" <aqwh54-NOSPAMPLEASE-@dsl.pipex.com> hath wroth:
>I'm hoping to provide 2x Load Balanced 8 Mbps adsl lines with a router that
>can carry out the load balancing. This would give a user more bandwidth than
>they would get on thier own ADSL line (providing that not many users are
>online at any one time)
That's probably overkill but will not solve the bandwidth management
and monitoring problem. My (semi-serious) rule of thumb on user
loading per T1 equivalent (about 1.5Mbits/sec) is:
100 light web and email users
10 business users doing whatever business users do.
1 file sharing user.
There's a bit more truth to the above than is obvious. Most (if not
all) peer to peer file sharing programs are designed to optimize their
use of the available bandwidth. Well, that's the official line. More
correctly, most file sharing programs are designed to dominate the
backhaul and monopolize all the available bandwidth. Note that the
typical ISP shows anywhere from 30% to 50% of their bandwidth used by
peer to peer file sharing. Your utilization will probably be similar.
One ISP didn't have any bandwidth management in place on their
internal LAN. One employee fired up some Bitorrent client and
successfully saturated a fractional OC-3 at about 50Mbits/sec. A clue
that something was wrong was a 100 times increase in the number of
support phone calls.
Incidentally, I've used several Edimax load balancing router models
with varying success:
http://www.edimax.com/html/english/p...-PRIrouter.htm
>They would also get the benifit that they would no longer have to purchase
>broadband and could in theory get rid of thier phone lines.
Yep. VoIP is cool. However, watch your pennies. If you hire a VoIP
to PSTN service, it can end up costing about the same as a single
phone line. Where it really saves money is with multiple phone lines.
As for quality, we're back to QoS and bandwidth management again.
>This would make
>this alternative an attractive one. My target is users who live on
>houseboats in a marina. There would be distinct advantages from being free
>from a copper line. They would have access when they have to move thier
>boats. They could save money if they don;t pay for broadband and a phone
>line.
Well, yeah. It makes sense and is probably sellable. However, be
advised that there are existing companies that sell similar marine
Wi-Fi services. You might wanna look at their costs and price
structure before jumping in with both feet.
>I've only looked into this because friends of mine live in the marina and
>they would like wireless internet.
Ok, there is your market research. Have you asked them how much they
are willing to pay you for the exercise? I've had similar unrealistic
expectations from my neighborhood WLAN/LAN. Everyone just assumed I
would do it all for free. I'm not sure about your cost structure, but
two 6Mbit/sec ADSL lines will cost about $130/month (after the first
years intro pricing). You didn't mention how many users, but I'll
guess you need about 10 to make this work. That's $13/month minimum
which is about half a residential phone line (after taxes). That's
assuming everything else is free including VoIP service and your
support services. Do you have 10 users that will pay even $13/month
for the backhaul on a regular basis? If your marina is infested with
the same breed of tighwads as found in the local harbor, methinks
you'll be lucky to get the $13 from any of them.
>I would like to think that I could deliver a reliable service and that the
>customers could actually save a bit of money in the process. Surely this is
>not a bad thing in the long run??
That's a good goal, especially the reliability part. The problem is
that your idea of deliverable reliability may not be quite the same as
their expectations. They probably want carrier grade reliability that
they have become accustomed to getting from their telco provider. The
first time someone fires up a leaky microwave oven and takes down your
entire network, they will be sure to complain.
As I previously mentioned and you seem to be ignoring, the technical
aspects are trivial compared to the social and political aspects. Look
again down my list of rhetorical questions and see if you have any
answers. If you think you have answers, then ask the same questions
of your customers to see if they have any expectations. If they don't
match, you're going to have problems. The rest is to just buy some
decent hardware, install it in a good location, pay for the backhaul,
and spend your spare time answering support questions.
--
Jeff Liebermann
jeffl@comix.santa-cruz.ca.us
150 Felker St #D
http://www.LearnByDestroying.com
Santa Cruz CA 95060
http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558