Second, the 60GHz links from Harmonix cost about $20000 for a pair of endpoints, and I bet the very similar Telaxis system costs about the same. (XXX -- new link, XXX, flog 802.16 even though i think standards for p-t-p radios are dumb). Since each rooftop unit contains four endpoints, this is too expensive. Radiant claims their cost is under EUR10,000 per node right now, and they're trying to reach $1100/subscriber. OptiMesh nodes cost about $20000 per roof including the physical node, installation, and support equipment, and each node contains 2 to 4 independently-aimable tranceivers, so adding 60GHz would double to triple the per-roof cost. I guess that's not really terrible, but it's pretty bad. Third, while I agree it's neat that OptiMesh rooftop units support arbitrary-medium same-speed backup links, I don't think the combination of Optimesh and fixed 60GHz dishes is a well-integrated system compared to either OptiMesh or Radiant's units operating alone. Turning a MAN mesh into a polished product is a big challenge, and both Airfiber and Radiant Networks have marketing that convinces me they've delivered serious contenders, in their respective physical mediums, but not when one system tries to cross into its competitor's medium. Will the OptiMesh system preserve its ease-of-installation if it uses Harmonix or Telaxis 60GHz as a backup? I think it will get much worse.
For example, the 60GHz antennas don't have that neat antenna-rotation feature that's so cool about Radiant's and Airfiber's rooftop units. The network management system is presumably logically dependent on this capability. Requiring techs to go to rooftops to resurvey and reaim links by hand gives the system a completely different character, and makes it less distinguishable from the readily-available sea of other radio alternatives.
There's other smaller problems, too. You'd have a lot of cables to organize. Airfiber's and Radiant's rooftop units working alone don't have these problems. Another integration problem is that to measure the 60GHz link's fade margin you would probably have to use the Telaxis or Harmonix management tool, not the OptiMesh tool, which is a big deal if the system offers reasonable centralized in-band management with logs and alarms and such. Sure, Airfiber could add some management gizmos, but I don't think they'll match (and maintain) all the feechurs in their partner company's management tool as well as they maintain the feechurs for managing their own hardware. There are probably other integration issues besides these.
Third, I don't think the backup link is necessary. For example, Telaxis claims their equipment, although rain-vulnerable, will achieve 99.999% reliability over 0.8km links in Seattle. In that case, why use the laser links at all? Why not use the OptiMesh rooftop ATM switches with the 60GHz link alone? Unless the laser is almost free, I think the extra link is a waste. I mean, get real. It's not like wired phone lines don't go out in severe storms---they absolutely do. so down with the Sieg Heil and take off your fucking Bellcore Armband, and accept the statistical argument.