The Rough Edge

Sunday, May 27 2018

Many of the technology trends over the last 40 or so years have been cyclical. One cycle that is often repeated is the massive move back-and-forth of where things are computed. Then pendulum swings between centralized and more distributed arrangements. In the early years of computing within a company, employees would ‘dumb’ terminals on their desks that connected to, and shared computing time from, a central computer (then called mainframes). As computers got faster and cheaper, much of the number crunching and day to day computing could be handled by machines at each person’s desk without off-loading too much to central servers.

There has been a big push over the last 5 to 8 years to move a lot of the data and processing that we interface with to the cloud so that precious electrons in our mobile phone batteries are reserved for more important things... at least that is the pitch.

Many of the things that we use the cloud for would could be handled just fine by a phone and would probably actually save a lot of those precious electrons. The amount of power used to run the cell or WiFi radio and crank out a bright back-lit screen that competes with the light is probably around the energy budget of running the super-secret algorithm right there in the palm of your hand. But that is just it, their super-secret algorithm would be free in the world and that would just not make financial sense to them. The cloud is not nearly as necessary for absolutely every company as it is sometimes made out to but it is easier in many ways to monetize.

The pendulum has started to swing back as we see large companies signaling that they intend to put more computing power at the edge. This seems to be in response to company policies requiring proprietary data live on premise and so the edge usually means some form of an on-the-premise, hybrid, solution with cloud components that are bolted on. There is nothing wrong with that - it is just not really ‘The Edge”.

Environments that are sometimes being touted as ‘The Edge” by many are still massive data-centers in relation to what we call ‘The Edge”.

The edge to us is processing data in the middle of nowhere on solar power and having to relay signals across miles of desert floor just to get to the very literal edge of where cell signals reach; and then only at 2G speeds!! Only 2G you say?? For us 2G is a 20 lane highway! When you are computing locally, you don’t have to be chatty with a server and you don't need a super highway. Everything has already been computed and compressed (by the power of the sun no less) and is pushed straight to the user’s mobile phone. If you are nearby then you won't even need the cell link. You can connect directly using only the local radios.

In short, real edge computing is not easy but we know it can be done.