Topic: Networking Questions (Read 1001 times)

Pretty simple, actually. Cat6 Ethernet cables can be run up to 100 meters (or 330 ft) without needing some form of extension. To extend it, you'd need to hit another network device (the simplest being a "repeater," which just re-powers the signal and sends it down the line.) Cat6 is the cable of choice because it provides more insulation and is less prone to crosstalk (interference.) The cable could be buried underground by itself, but it would be safer to run some PVC pipe just to provide some more protection.

I have a friend who is very interested in mesh networks. I think he sees it as a way to be independent of mobile carriers.

However my gut feeling is that you won't be able to beat the carriers in any dimension except non-carrier-ness with todays technology. The way 3G/LTE networks are set up is pretty damn efficient. These networks are carefully optimized by professional engineers who spend a lot of time doing capacity planning and tuning. Data on these networks takes very direct routes. If you sit in the middle of London and ping a New York IP address from a phone your data travels from your phone to a nearby tower, probably then to a local switching center via fibre or microwave link, then almost immediately hits the inter-ocean links and arrives at its destination. All these links are carefully monitored to avoid congestion and trigger rapid repairs if they are damaged. Overheads are minimal because if somebody starts abusing the network they can simply be cut off.

A mesh network has a lot of unsolved problems to deal with. One is latency. Unless your mesh nodes cover wide areas there might be many hops to reach an internet access point. Each hop adds latency due to radio delays, routing etc. If the node hardware is cheap it'll just be even worse. Think Tor vs direct connections.

Another is abuse, deliberate or accidental. If somebody comes up with a home-grown protocol stack implementation and starts spamming the airwaves with bad packets, existing mobile networks can just shut those people off. There's a clear hierarchy of authority and the networks are secured with mutual network/handset authentication. In a mesh network nobody is really in charge so there's nobody who can make the "kill this device" call if something goes wrong. People who are being antisocial could not be stopped and the result would likely be blacklisting of the meshes internet bridge points.

Yet another is routing. Even if you can tolerate the latency and there's no buggy/abusive nodes, finding a path through the network that isn't overloaded and is fast would be hard. The internet at the backbone level is a mesh network and manages this using BGP, but it requires expensive hardware and the constant vigilance of trained operators. BGP does not address load balancing across links for example.

In short I don't think anyone will be interested in a mesh network any time soon for the same sorts of reasons that people don't usually browse through Tor by default. Small ad-hoc meshes might be useful in specialized scenarios, like towns in the middle of the desert.

I have a bunch of questions about general networking, if you guys could answer some it would be very much appreciated. Most of it is about decentralized networking.

1. How viable is a world wide wireless mesh network (obviously you still need a cable to go to and from other continents.)
how much throughput and what would the ping be like talking to nodes across countries like the US. How well does it scale, and overall how viable is it, and how much would it cost.. what are the disadvantages.

2. How expensive is this http://en.wikipedia.org/wiki/Free-space_optical_communication . Would it be good for #1 and the wireless be a backup so to say when conditions are not good for FSO.

3. Would you rather pay anonymously for a mesh network like this or continue paying your current provider.

4. You and your neighbors have to chip in some money to buy good equipment for your street/block/complex/flats. How would setting up a cable across long expanses of land work?

5. How easy would it be to exploit these types of networks.

Thanks for any answers.