Who is most reliable vendor for outdoor Point-to-Point?

Fiber's going to be about your only option. WiFi isn't nearly that reliable, and the interference sources moving into your neighborhood 5 years from now will trash your existing wireless network.

Care to share the details of your equipment, network, and reliability requirements?

You'll probably want to talk to the companies that work in industrial data collection. The two market leaders are:

In a previous position I installed both Symbol and Intermec equipment outside at a ski resort for wireless data applications. Provided you bought their ruggedized (read: expensive) gear, it worked well.

Keep in mind anything outside is going to need some TLC occasionally, i.e. to brush off the snow and ICE, clear leaves or what have you.

Cheers, Geoff Glave Vancouver, Canada

If you want reliability, the telecom vendors found the method years ago. Spatial and frequency diversity. Basically, it's two radios, in two parts of the tower, on different frequencies, with fail-over protection. If something trashes one frequency, the other is likely to still be working. If some reflective, diffractive, or absorbent object gets in the way of one path, the other is highly likely to still be functional. That also includes separate power sources and redundant data paths. Redundancy also allows for preventive maintenance of one radio, while the other continues to shovel traffic. To insure that things don't deteriorate, drift, or just plain fail, monitoring is mandatory. That means some kind of SCADA system with a data logger and threshold alarms to alert the maintenance people that something is about to blow up.

I forgot to mention that bullet holes in the antennas was a problem with fiberglass that tended to shatter. Panel antennas were far too easy for target practice. I suggest spun aluminum dish antennas with pressurized (and alarmed) Heliax. At least you know when someone has shot a hole in the antenna.

Methinks your 10 years of uptime is unrealistic for a simple reason. The error rate in modern DRAM and uP systems is sufficiently high to insure at least one crash per year (or more). If the inherent soft error rate doesn't crash the box, cosmic rays will also do the same thing.

RAM and redundant multiprocessors are a big improvement for servers, but have not been used in common wireless equipment. More commonly, the radios have a totally independent watchdog timer to verify that nothing has crashed. (Note: software watchdog timers don't work well). I would not expect anything to stay up for 10 years, but with regular reboots, redundancy, monitoring, and preventive maintenance, it will last for 10 years.

Incidentally, I have some Breezecom Alvarion AP-10/SA-10 2.4GHz links running at 3Mbit/sec (1.2Mbits/sec thruput) that have been up since about 1994 and are working just fine. They have to be remotely reset on occasion, but have never failed (except when someone plugged the wrong power supply into one and blew up the protection diode). Biggest headache are the trees that grew into the line of sight.

No comment. All I know is the distance and a very unrealistic reliability figure. No clue on speed, cost limitations, environment, and topology. How about a better clue as to what you're doing and what's connected on each end?

What OS will you use? Certainly not anything from uSoft.

It varies depending on complexity, altitude, and chip density.

A cell phone with one 4-Mbit, low-power memory with an SER of 1000 FITs per megabit will likely have a soft error every 28 years. A high-end router with 10 Gbits of SRAM and an SER of 600 FITs per megabit can experience an error every 170 hours. For a router farm that uses 100 Gbits of memory, a potential networking error interrupting its proper operation could occur every 17 hours. Finally, consider a person on an airplane over the Atlantic at 35,000 ft working on a laptop with 256 Mbytes (2 Gbits) of memory. At this altitude, the SER of 600 FITs per megabit becomes 100,000 FITs per megabit, resulting in a potential error every five hours. The FIT rate of soft errors is more than 10 times the typical FIT rate for a hard reliability failure. Soft errors are not the same concern for cell phones as they can be for systems using a large amount of memory.

More numbers at the bottom of:

in: I've kept servers up for more than a year. Some servers I've seen show uptimes of perhaps 2 years. I don't think any device can stay up without rebooting for 10 years, especially if updates and cleaning are required.

With all due respect to both you and Wikipedia, even commodity microprocessor-based systems are roughly an order of magnitude better than that. Cosmic rays are very old news.