ExtremeTech has an article covering the Intel Developer Forum which, as always, includes Intel’s vision of the future.
It comes as no surprise that Intel’s visions are interesting, sometimes bizarre, and sometimes mind-blowing (it comes with the silicon territory). Today, Intel predicted that by the year 2020, the energy cost of doing meaningful computing will approach zero. What impact does that have on Silicon Valley, and how will those ripples effect the world?
Intel is famous for prognostication. They’ve been at the forefront of the technical world since the early days of Silicon Valley, largely due to their knowledge and expertise in silicon fabrication. Now, Datacenters are less about raw computational power and more about using less energy to accomplish the same tasks. Hell, they’ve gone as far as submerging their servers in liquid pools as a means of reducing the costs of power. But why?
It’s no wonder since Intel’s focus has been on power consumption for quite sometime, but it is important to understand the ramifications of their prediction. If meaningful computing has no significant energy cost, computers performing valuable tasks will become ubiquitous. Right now, ubiquity extends to cellular communications, but, as Cisco’s Internet of Things would have you believe, the world will change as sensors cover the globe. Weather, right now, suffers from inadequate measurement, but companies like climate.com are changing this paradigm by building their own sensory networks.
Today, the quantity of machine data that is produced far exceeds the quantity of human data (the scale probably tipped around 2004). This explosion in machine data allows us to analyze and understand information in new and interesting ways, but the infrastructure to support these kinds of developments is still in the pipeline. We need analytical software for digesting this data, and Internet connections to move this data, and switches to route this data, and all of these wonderful forms of infrastructure need orchestration.
Assuming Intel is right about the cost of powering computing decreasing, all that’s needed for Cisco’s vision to come true is some advances in material sciences and battery life. The big issue with Billions and Billions of sensors is administration. Put simply: who’s going to change all those batteries or perform maintenance? The answer: no one.
New sensors will have to be self-powered, biodegradable, resistant to wear and cheap! All four requirements will have to come to pass before sensors can approach ubiquity (which, going back to my earlier discussion of cellphones, is not 1:1 with humans. Sensor saturation is likely many multiples of the number of cells we have now).
Intel’s vision today, whether it comes to pass by 2020 or not, is a necessary component of the next big network. Make no mistake, in 8 years, machines not humans, will be the largest consumers of network capacity. Service layers, beyond the physical requirements, will have to spring up to support this world, and there will be many new technologies. The future isn’t a cloud, but a web.
Disclaimer: http://2600hz.com, where I work, thinks a lot about The Internet of Things. We might have some plans around this vision.