Shaping up the new Cloud

The evolving IoT Edge cloud will ultimately surpass the one that gave us smartphones and tablets. It will also spur dramatic technology growth.

By Kenneth P. Birman

The convergence of 5G networking with advances in artificial intelligence (AI), sensor technologies, robotics, and mobility has created an immense business opportunity. It will not be long before we see a new form of the cloud: the internet of things (IoT) Edge cloud that may ultimately surpass the one that gave us our smartphones and tablets. This IoT cloud revolution will also spur dramatic technology growth.

During the 1990s, the first web boom, I attended a small summit at which industry leaders debated the key factors that underlay technology disruption. One of the speakers asked why the internet had been so disruptive and offered the following hypothesis. In the 1980s Moore’s Law enabled steady advances in computer hardware: displays grew in size, pixel density, and began to support color. DRAM capacity and speed increased, storage capacity and speeds climbed, and clock rates advanced. Inside our companies, we had Ethernet running at 1Mbps. Yet for all its technological wonders, this period of growth was remarkably balanced and not hugely disruptive. It gave us increasingly powerful computers and enabled very effective enterprise computing systems, but the experience wasn’t radically different from a few years earlier.

Today, the uplink path from IoT sensors to the cloud is so slow that it blocks us from integrating those devices into the cloud in a direct, real-time manner.

What none of us was thinking about was the sense in which the wider networking options were limiting progress. In those days, there was no true public internet, so most communications remained sluggish. Even the most powerful computers used dial-up modems over telephone lines to communicate at a geographic scale, and in fact, this was also the WAN link technology for the nascent internet itself.

But then the communications sector experienced its own revolution. Enterprise networks upgraded to switched optical, and our routers could forward packets directly over a worldwide internet at lines speeds. Suddenly, all the ingredients for explosive growth fell into place. Overnight, a web revolution swept the planet. An immense disruption… and yet, even when viewed back in the 1990s, one that should have been completely predictable.

Today, we stand at the threshold of another such revolution. Of course, the bottled-up technologies are not identical. We possess a diverse collection of sensing devices, cameras, videos, microphones, and all the “things” in IoT and IIoT, a cost-effective cloud computing model, and powerful AI-enabled by big data and massive hardware accelerators that run on that cloud. Yet much as in the 1990s, the true opportunity is limited by a daunting communication bottleneck: the uplink path from IoT sensors to the cloud is so slow that it blocks us from integrating those devices into the cloud in a direct, real-time manner.

The intelligent IoT edge cloud will ignite the same sort of explosive growth that brought us modern data centers. Trillion-dollar industries will quickly follow.

Consider image-based intelligence, which is the key to many of the most exciting opportunities. The images are just too big to upload, and this is so even for photos – and far worse for high-resolution video. The industry is trying to do more and more computing on the sensors themselves. But even the smartest cameras are constrained by a lack of resources and electric power and lack the situational perspective that can be achieved only by synthesizing data flows across thousands of sources. So, what is the bottom line? Without the big data capacity of the cloud and the hardware accelerators, one simply cannot tap into the most exciting AI tools. Any application that involves dynamic machine learning (ML), in real-time, or coordinated reactions to unpredictable conditions, is limited by this communications barrier.

Dam up a huge river and the waters behind it will rise until they overtop the dam and wash it away in a massive flood. In the 1990s, that dam centered on dial-up modems, and we tore it down by enabling a true high-speed, wide-area internet. Today’s IoT and IIoT barrier involves networking at the edge, and this time, the disruptor will be 5G networking.

With 5G, we add two new elements to the mix. First, 5G networks are blazingly fast: once the 5G rollout gets serious traction, speeds of 10Gbps to 20Gbps should become commonplace. We’ll suddenly be in a position to upload all of that sensory data. This leads to the second development: 5G networks situate small compute clusters on the access points, giving us a place to host a new kind of cloud. By treating these small cloudlets as a new form of cloud, with their own platform-as-a-service solutions and its own hosting options, developers can create applications to leverage that vast stream of edge information. Moreover, by keeping the data close to where we capture it and then discarding it promptly, we can even provide privacy guarantees. The intelligent IoT edge cloud will ignite the same sort of explosive growth that brought us modern data centers. Trillion-dollar industries will quickly follow.

Research efforts are also on to invent the software enabler for that next step and create a new hosting platform that can help modern ML and AI solutions run close to the devices, connected over 5G. This will enable a true synthesis of IoT/IIoT with federated AI. The power of the cloud can expand to encompass the IoT edge.

An intelligent IoT Edge is also coming soon, and it will transform the world, giving us genuinely smart homes, smart power grids that can better leverage renewables, smart traffic intersections, roadways and cities, and even smart farms. AI will slash energy waste, reduce farm runoffs, and enable us to treat insect and other crop maladies with pinpoint precision, using far less pesticide to far greater effect. And the technology leaders who create this new infrastructure will lead in the transformation of today’s communications industry into tomorrow’s most vibrant and fastest-expanding cloud.

Birman is the N. Rama Rao Professor of Computer Science, Cornell University, where he leads the Derecho and Cascade projects, creating new technology for the IoT Edge Cloud

feedbackvnd@cybermedia.co.in