As technology reminds us, not all trends follow the same path. While the press like to announce that public cloud is dominating compute and storage and that the remaining holdout platforms continue to transition from on-premise in favor of these same public clouds. But not all compute exists as SaaS or PaaS, running on IaaS, and not by a long shot. Today (2018) over 2 billion people have a smart phone worldwide. That means that 2 in 7 people on the planet have a phone with significant compute and storage, and being a smartphone means people are more likely than not to store media and content that is not necessarily in the cloud. This percentage is expected to grow and at rate that parallels the increases in smartphone SOC innovation. Smartphone buyers are not typically one-and-done buyers. This explosive growth in consumer electronics in our pockets is an untapped form of cloud compute, one that is going to change how technologists think about the edge.
As compute power has evolved, so has the need for high speed broadband to meet changing consumers’ expectations – to parallel pop culture references like if you build they will come, but also when do we want it – now! New consumer demands like OTT video and IoT are increasing throughput demands on existing networks, but new devices are generating consumer electronic ripples that are likely to rewrite current experiences. A new type of content consumption, one that can be traced back to the 3D movie and video and more recently gaming and consoles. Multiple industries, that previously have had little interaction are now co-developing and investing billions into the next breakout device. What has been called the head mounted display or HMD, is promising to revolutionize the world, not just entertainment. These new immersive devices are limited more by our imagination than current technological capabilities. An entire new science for creating VR and AR content, physiological adaptations, new hardware, haptics, new screens, and cheap wireless are creating this new market. In order to both maximize and monetize the experience; these new technologies are blazing a trail. By immersing into the content instead of just watching, the traditional idea of visual effects are undergoing an adaptation to a new virtual and augmented world.
The primary limiting factor today preventing meaningful virtualization in the VFX world is latency. The laws of physics tell us data cannot travel faster than the speed of light between the core and the edge, therefore something else needs to give. One dominant idea is that compute will undergo a shift to split between the compute of image processing at the core and the rendering of images at the edge. The future of the CPE, the expectation and demand, whether a wearable, optical, a 2D wall of light, or even embedded implant; compute will need to increase to offset latency. By separating the compute intensive processing from the final image rendering, the data can be stripped down to a metadata stream that is a lightweight stream of parts, capable of representing the images in a way that can be assembled (rendered) independent of the bulky source, but at the edge (where the consumer is at that moment). This opens up the artist to move from the 2D desktop to the XYZ axis of the workspace.
Whether this is watching theatrical, episodic, user generated, or gaming, the traditional television is becoming obsolete. The debate at the moment is what is that edge compute device; is it the smart phone, the HMD, a console, a STB, or something new. What is likely to be true regardless is that edge compute is going to take lessons, technology and even architecture from the cloud. Today we are seeing containers, a traditional cloud and infrastructure technology, showing up in CPE middleware (OS), including Amazon’s own Lambda (GreenGrass) and LXC from Linux.