This blog post was originally published by Bitfury. It is reprinted here with the permission of Bitfury.
Shelter in place. Lockdown. Curfew.
Where I live, these terms are only heard if we have a bad Nor’easter snowstorm headed our way. But with COVID-19, these terms have now entered daily life as we watch this terrible virus sweep across the world. There are now strong international directives urging people to stay home in the hopes of flattening the curve, instructions on how best to wash your hands and how you can help your elderly and immunocompromised neighbors from needing to go out. Most researchers predict that these trends will last for months as we all adjust to our ‘new normal.’
In the midst of all of this, however, I think there are several opportunities for technology to help us emerge from this pandemic better prepared to provide human and social safety — both now, and in the future.
Let’s take a look at an increasingly popular use case — IoT-enabled home delivery — to see how this could work on a national or global scale.
Your Refrigerator’s Amazon Order is Here
In 2015, IoT for refrigerators was just coming into fashion. At the time, it seemed silly to me — I mean, who really needed an internet connected refrigerator to scan your food and tell you what you needed to purchase, and when? Or, even more far-fetched, would you really trust your refrigerator to order it for you?
Fast-forward to the discussion I had with my wife last night about our COVID-19 preparations. We decided to put together a 14-day list of meals and shopping list so that we will only leave our home once in these next 2 weeks to stock up on essential food, medicine and toiletries. (No, we aren’t going to purchase a year’s supply of toilet paper!)
This pattern of shopping is unusual for us, as we usually shop in a more European style, going to the store every few days. We were definitely not the target market for smart refrigerators or even grocery delivery services — mainly because it didn’t seem worth the money just to save on a few hours of grocery shopping. Now this decision has a whole new dimension — while it could save you time, it might also potentially protect you from infection.
Recently, a colleague of mine said that many people are building new homes with small interior/exterior rooms that feature remote security access and refrigeration. The idea is that this allows for grocery deliveries even when you’re not home. Now equip this scenario with pandemic preparedness in mind. We could design these rooms to accept deliveries even while we are confined to our houses, keeping ourselves and the delivery drivers safe, all by using IoT cameras, locks and refrigerators.
Can the Internet handle massive adoption of IoT?
Another aspect of this trend is whether it’s even possible from the standpoint of internet bandwidth. Let’s look at another example to explore this issue — IoT security devices.
I’ve recently considered purchasing IoT-enabled outdoor flood lights. These devices use high-resolution cameras to watch for movement and trigger light to provide increased security to your home. These cameras are incredibly advanced — they can distinguish from family members/frequent visitors and issue advanced alerts if someone isn’t recognized. To do this, these devices need to make computations and decisions quickly. The best way to do this is to have the camera complement onboard compute by communicating with a “local” data center, as it’s much faster for a camera in New Jersey to speak to a data center in New Jersey than to send data across the country to Nebraska for computation, and then back. By using a local data center (a.k.a. an “edge data center”), we reduce the amount of bandwidth, cloud processing and cloud storage needed by these devices, and we can return computational results much faster.
Here’s the problem — there aren’t many edge data centers yet. Many organizations rely on regional data centers, so your IoT camera in Arizona could be speaking to a data center in Colorado (along with the cameras from the surrounding states of Utah, Wyoming and Kansas). That’s a lot of computation to happen at one location and will often result in slower compute times — especially in a time of increasing internet usage. We’re already seeing our internet networks become stressed by the COVID-19 pandemic, with more people working at home, schooling at home and binge-watching Netflix all day. Our networks simply weren’t designed for this load — and now we’re asking them to handle trillions of IoT devices as well.
About Edge Computing
Edge computing is essentially a web of internet-connected devices that offer high performance computing, either on the device or via edge data centers. The key to IoT and edge computing is that these devices do more onboard computing and data storage to reduce latency for critical applications. Having the ability to process and analyze data locally, before sending it to a regional or centralized data center, can realize massive savings in network bandwidth and cellular access costs, as well as increased computation speeds.
Back to your IoT food delivery: The next generation of IoT refrigerators will likely use a distributed computing platform which maximizes the ability of technology to make our lives both simpler and safer. Coupled with 5G, the experience of planning your food purchases may come down to a one-time setting of your food and beverage profile, followed by advanced learning and AI facilitated by IoT devices, which are connected to edge data centers to execute quickly. Your Amazon order is on the way!
However, the solution isn’t just to build more traditional data centers. Edge data centers will be much smaller and located in places we could have never imagined. The proliferation of these sites will lead to increased energy usage. If designed incorrectly, this could be seriously unsustainable.
Building for Edge IoT
Edge computing began with the rise of centralized processing and hyperscale data centers. But as machine learning, AI and edge computing have evolved, divergent processor architectures have converged to provide the most efficient assignments of compute workloads, thereby improving performance and power. However, the use of higher-powered CPUs, GPUs, ASICs and DSPs challenges the practical limitations of air-cooling — read more here about the trade-offs between sustainability and efficiency in data centers.
At Bitfury Allied Control, we are focused on finding the best ways to cool edge and micro data centers. We have developed 2-phase liquid cooled immersion systems which allow edge computing to be compacted in about half the amount of space compared to an equivalent air-cooled system. This is especially important, as the edge will require hundreds (if not thousands) of sites across a square city kilometer (think street corners, light posts, etc.). Bitfury Allied Control immersive cooling has excellent energy efficiency, operates without noise and is impervious to dust, particulates, liquids or contaminants, making it ideal for edge data centers.
However, we know that we can’t do this alone. The surge of demand for IoT devices and edge computing will require the entire IT community to work together. By doing so, we can help preserve the security, health and quality of life for users all over the world. If you’re working on edge computing applications and want to know more about liquid cooling, please get in touch with me at firstname.lastname@example.org — we’re ready to help.
There is no doubt that the devices we use every day are becoming more intelligent and more powerful. While initially designed to help simplify our lives, I truly believe they have the potential to make our lives safer as well.
As we connect more devices to the internet, and more devices at the edge, centralized data centers will need to be complemented by edge and micro data centers. This will allow us to leverage the cloud to drive a virtually limitless number of new IoT applications — meaning you can rest-assured your smart fridge or IoT camera are working at their peak performance and keeping you safe.
Learn more at https://www.allied-control.com/.
Head of Bitfury Allied Control