Robert Zemeckis and Bob Gale were right about a lot of things in their eagerly anticipated 1989 sequel Back to the Future Part II. Much of the movie was set in then distant future, 2015, where science fiction had merged with our everyday lives in some remarkable ways. The big question is – how many of the predictions actually came to pass?
Against all the odds
We did indeed embrace the practice of video-calling, a notion that not too long ago seemed like it just wouldn’t work, less for technical reasons and more for social ones – who wants to have to comb their hair and look presentable just to ring their mates? Not only has this premonition become almost ubiquitous among friends and families with OTT mobile apps like Skype and WhatsApp, but with the growth of social media stories and live streaming platforms, many of us seem more and more willing to broadcast ourselves to the entire world, sharing our deepest thoughts, opinions and brunch plates to all who care to listen and consume.
When fiction becomes reality
Robert and Bob were also spot on about the evolution of video games. Think of the moment, when Marty McFly is appalled to learn that the “kids of 2015” are only mildly amused by the cutting-edge video arcade game “Wild Gunman”, referring to it as “like a baby’s toy!” When we look at the virtual reality, motion-enabled games being played today, those plastic 6-shooters and 8-bit graphics really do seem a little fundamental – although undoubtedly still a lot of fun!
And the not so accurate predictions
However, some predictions were a little over zealous. Perhaps, the most famous and daring prediction came in the opening sequence, “Where we’re going, we don’t need roads.” This impactful and dramatic statement as the DeLorean shoots off into the sky reminds me of some claims and rhetoric I hear regularly in the Tech industry along the lines of, “Hardware is just a necessary evil, it’s a software defined world and that’s that.”
We need software and hardware
I’m not going to argue that the present and future of our industry is not significantly defined by software. The benefits of modern Cloud Architecture, Virtualization, Containerization and advances in Management/Orchestration are certainly not going away. Indeed, they will be the technologies that will support the next generation of software innovators and OEMs to bring products to new markets at a scale and velocity, previously unheard of.
However, despite this, the hardware elements of the solutions that we put together continues to be critical. In fact, I believe that hardware infrastructure is becoming more and more important– effectively, enabling the software “DeLoreans” to move faster, more securely, more efficiently and into new areas of our world. In my view, there has never been a more exciting time to be involved in hardware solutions.
At the Edge
After all, Edge computing means that we have computing resources, applications and data-analytics closer to the point of data generation than ever before. For me, “Edge” translates into decentralized as “not in the core datacenter.” It is represented by computing in cell towers, retail outlets and power substations. Edge means harsh conditions, small spaces, limited power and cooling. Edge equates to physical security, smart remote management and modularity.
Bringing light to the darker corners
These considerations surely give rise to important software considerations, but at its very core (pun intended), Edge computing demands continued innovation in hardware design to bring new light to the darker corners of the network. Without this innovation, we will be unable to realize the dreams that inspire the software visionaries to write the future.
Domain specific hardware architectures
On that note, domain specific hardware architectures are already enabling massively accelerated performance for software functions that require specific types of computational mathematics. Picture FPGA based accelerators from companies like Intel and Xilinx, validated in Dell Technologies Servers that can accelerate low-precision floating point calculations with extremely low latency.
Likewise, for activities like machine learning model training, GPUs such as NVIDIA’s V100 offer huge performance gains versus more traditional, less parallel architectures. Accelerating Machine Learning algorithms is becoming an increasingly competitive market, with new purpose-built hardware architectures being designed that are optimized for the convolutional mathematics used within. Smart NICs are offloading network functions to optimize IO while purpose-specific silicon is enabling new root-of-trust functionality for securing applications at a hardware level. And, let’s not forget cryptographic security, where keys are being stored and generated on hardware subsystems.
The holistic view
In short, your software can be amazing, but if the hardware platform is sub-optimal, it can be hobbled into something slow, insecure and unreliable. Looking holistically at the whole solution, across both hardware and software, is the key to delivering truly amazing products. If you’re trivializing either element, when developing your go-to-market strategy for your product, you will fall short of excellence.
The good news is that we have decades of experience in designing products both from a hardware and software perspective that enable innovators across multiple industries to leverage the best emerging technology for their applications. Why? Because even when we finally get our hoverboards – we still need somewhere to put our feet!