This is indeed an interesting topic and something I
touched on before. I think there is a lot to be learned from the mobile networks in this regard - one of my first jobs 10 or 15 years ago was working on the design of Australia's first digital mobile network and aside from coverage prediction (which would have been much easier with EC2 rather than manually pushing raw data from Sydney to Melbourne for processing) we would look at how the base stations would be connected to the network, where they would be located, what frequencies they would use and so on. Many of the same problems (security, power, cooling, monitoring, connectivity, leasing, etc.) have been solved time and time again, and I'm sure that there are lessons to be learnt from this existing work. Mobile providers would indeed be very well placed to deliver this type of edge-based architecture already, at least where traditional base stations/
macrocells rather than
microcells (or smaller) are used. They have many well connected, powered, secured and cooled locations in which housing a rack of forty-something servers should be no problem; they could even look at using liquid cooling and selling the heated water back to the surrounding building(s) for heating and/or industrial use.
In fact they are such similar areas that
prior art like
this report by one of our competitors at the time (and
more recent ones by suppliers like Ericsson) on the subject of stuffing shipping containers with computerised base station hardware could well scuttle attempts to
patent the same thing in the context of cloud computing.
Cheers,
Sam