Preparing For War On The Edge

War clouds are gathering over the edge of the network.

The rush by the reigning giants of data—IBM, Amazon, Facebook, Alibaba, Baidu, Microsoft and Apple—to control the cloud by building mammoth hyperscale data centers  is being met with uncertainty at the edge of the network. In fact, just the emergence of the edge could mean that all bets are off when it comes to data dominance.

It’s not that the big cloud operations are in danger of losing business. But they may be in danger of losing control over some or all of the most valuable data, and that could have significant repercussions on a global level. So what will they or others do about these changes? At this point, it’s not clear how all of this will play out.

The edge, which even a year ago was largely ignored by investors and chipmakers, has suddenly become the next big thing. Companies are putting labels on it, defining its boundaries, and weaving marketing messages about how all of the pieces will fit together. No one knows yet if any of that is correct. What is clear is there is too much data being produced by edge devices, and it’s too costly to send it all to the cloud. In safety-critical applications such as autonomous vehicles, waiting for a response from the cloud could be life-threatening. The higher the speed at which a vehicle travels, the faster the response needs to be, and the cloud—even with a millimeter-wave 5G connection, isn’t fast enough.

Big server and chip companies are pitching a hybrid cloud solution. The general idea is that some data will be processed inside of devices, whether that’s a car or a factory or a smart watch, while other data will be processed in edge clouds, with less-critical data processed and stored in the cloud. It’s a coherent vision for the largest tech companies, but no one knows yet if it’s correct.

While there certainly will be a need for clouds, it’s not clear what kind of data they will be storing, where they will store it, or who will own that data. That storage may use IBM or Intel chips, but it also could use Arm, MIPS, or RISC-V cores, FPGAs, or any variety of other processing elements. In fact, the key to this shift may be less about a single solution than more customized solutions based on domain expertise.

In automotive, for example, the large OEMs may opt for their own cloud infrastructure and differentiate based upon the response time for a variety of non-critical services. And in the industrial world, clouds could well be on-premises using customized hardware built from a variety of chiplets and packaged together with unique sensors for vibration, noise, odor, movement, sound, heat and streaming images.

With on-site analytics, that kind of data can add a competitive edge for companies and much faster response time to deal with any aberrant data. So rather than waiting for data to be processed, stored and retrieved by the hyperscale cloud, problems can be solved in real-time using machine learning and localized data. Or even better, data can be monitored in real-time using predictive analytics to optimize uptime and productivity.

There are still many uses for the massive horsepower in clouds. These are basically supercomputers for hire, and in EDA, material exploration and just enormous number crunching, having that kind of compute horsepower is extremely valuable. But plans to centralize all data forever are looking increasingly shaky as the edge begins to take hold, which is why all of the big cloud companies are suddenly getting so active on the edge.

The big question now is whether they can extend their dominance from the cloud to the edge, or whether they’ll be pre-empted by an army of domain experts with better ideas about how to leverage their own data.