Bob McCormick W1QA wrote: > In what I've drawn up so far - one of the weakest points > in our whole configuration is the D-STAR repeater controller > itself! Its easy to configure redundancy in that last mile > network connection ... and setup a reliable gateway including > maybe even a backup system. We can sustain the failure of > a band module - which would leave other modules available. > But if we lose the controller itself ... ugh!
Bob, this is actually a very interesting topic. One thing I've heard some groups have done is bypass that "power bus" topology of the controller altogether. It really does nothing more than turn power distribution into a single-point-of-failure. Plugging the modules into power directly would not (as far as I can tell) suffer any ill effects, unless they were accidentally hooked to floating grounds that didn't tie together. That could make the serial protocol between the controller and the modules unhappy if current were flowing. The "brains" of the controller is harder. One could buy two of them and build a complex switching system for the four module cables and the Ethernet... but that device probably introduces more possible points of failure than the controller itself does. Hmm... interesting topic. There's also been some discussion about how to "harden" the Gateway server properly, but those are generally well-known best-practices that any server admin who understands Linux could accomplish pretty easily. Same with the network gear, kinda... at least there's some commercial class networking gear that could be pressed into service and probably never show a fault in many years of operating time. Even some "semi-commercial" quality gear will do that nowadays. What else would need to be considered? Obviously a power source... automatic generator, if really paranoid a battery plant (and figure out some way to alert over D-STAR itself that the battery is online and site power is off... hmmmm... another side project...), etc. Nate WY0X