Edge computing offers enormous promise – some say it may even supplant cloud computing. Certainly this emerging technology, in which sensors across the Web provide a torrent of data, is growing rapidly. A research report in August 2019 forecast a blistering 32% CAGR yearly increase between now and 2023, meaning the edge market will double in size.
Edge computing fuels many of the tech trends that are getting buzz today, including smart factories, smart grids, connected vehicles and more. While IoT has driven edge computing, the technology – fueled by 5G – will play an ever greater role in many sectors beyond IoT.
To provide insight into the future of this key technology, I spoke with a leading expert, Bryan Beal, Senior Director, Strategy and Solution Innovation at VMware – Telco and Edge Cloud Group.
DOWNLOAD THE PODCAST:
See Transcribed highlights below:
1. What Exactly is Edge Computing? (4:11)
Beal: "So, I think the industry is still trying to decide on a very firm definition, it's kind of like cloud 12 year-plus years ago. We're still moving it around, but if you think of it, it's moving the compute and the use case as close to the user as possible, with the goal of being a better user experience for that application. So if you think of how cloud centralized everything into one big data center, the edge is basically the peeling back of that and actually moving stuff out closer to the user in a much smaller footprint. That's basically the short version.
"[Sensors are] one use case. It could be many things, it could be everything from cloud gaming to caching a video stream, similar to CDN, it could be 5G is enabling a lot of use cases that consume edge for things like network slicing, security, quality of service. It could even be a white box in our house that's provided by our TV or internet company to power things like our home phone service and our wifi. It could be many, many things, but effectively, it's getting a little bit of compute, a little bit of network and a little bit of storage as close to that user as possible.
"You can't break physics. And at the end of the day, the closer that the use case is to the user, the lower the latency is and for many use cases that gives a better user experience.
"But there's also tiers of edge. There's a concept of near edge, far edge, deep edge. Sometimes it's called different things, but if you look at a telco, they might have a large central data center, but then they also have these central offices all over, and they might consider that a near-edge location. And then you get farther and farther out to maybe a box on the street or in an apartment complex, and that's one step closer to the customer, so there's actually tiers to edge as you get from the central data center all the way out to the user."
2. Use Cases for Edge Computing, Edge vs. Cloud (1:59)
Beal: "I can tell you our customers are considering everything from IoT to... Gaming has become a big one, actually, there's a lot of focus on getting the experience for the kids coming home from school and popping open whatever the latest game is and maybe even being able to play with other kids in the neighborhood and having a server that could sit right in the neighborhood.
"So the kids from school are playing together with a really good experience with low latency, but there are many definitions.
"I think edge is going to create an entire new industry of startups and cool things, just like LTE did or 4G did for mobile applications.
"I don't see [edge] killing cloud at all. I see your analysis as being spot on. These are two complementary technologies. These are things where we'll use the edge for what we can and where latency matters, or where maybe we can limit the amount of back-haul traffic on the WAN back to a large data center.
"But it's not going to replace the cloud. At the end of the day, you get economies of scale from aggregating everything in one big place, so some workloads will always work better in a large cloud architecture; other workloads are gonna work better on a small chip sitting out on your smart meter or your toaster, for example."
3. What Really Captures your Imagination about Edge Computing? What about the Role of 5G?(5:24)
Beal: "It would be hard to pick one. 5G certainly would be a top of mind one right now, because you really can't do effective 5G without edge. Those two really kinda go hand in hand. Now, I can do edge without 5G, but 5G, especially millimeter wave 5G, very dense urban 5G, requires all these new radios, all these new sites that are gonna be put up.
"And the operators see that as an opportunity to have basically a mini-server in each of those sites to enable them to make money selling more services, like network slicing is certainly one example, where you can deliver really high reliability and low latency for a use case like an autonomous car, robotic surgery. And then you have another slice for things like your toaster, your IoT that probably doesn't need the same experience. And then you have a slice for the massive mobile broadband, speed, the fast slice. And so, I think 5G and edge and how those two are gonna come together is from my table one of the coolest things that I see.
"I think 2021 will be the breakout year for 5G, I think that's when we're finally gonna see the coming together of real network coverage, not just in certain little blocks in a downtown area, but broad network coverage of 5G. And then more devices that support it, you kind of need both. It's a chicken/egg thing. And what a lot of people don't realize, 5G is more than just millimeter wave in a downtown area, it can be used on the rural spectrum, the older 800 megahertz spectrum. So, I think we're a couple of years away, but I think... Yeah, it's probably 2021 is when we're really gonna start to see it hit mainstream.
"I see a couple of things [happening with edge, as 5G matures]. I see the operators being able to make a lot more money off of what they do with that edge footprint, and that could be a wireless operator on 5G, but even a Comcast, Charter, Spectrum, Cox, the cable companies. They have this great network as well, and it's really those service providers that I think are gonna leverage their footprint, their network, to enable all these new edge use cases, but I don't think they're going to be the only ultimate provider of whatever the cool thing is.
"The start-up community, right, just as we saw with 4G, brought us Uber and GrubHub and 20,000 other things and mobile gaming that wasn't really possible before 4G. I think edge plus 5G, or even edge alone, honestly, is going to bring a whole new industry of startups, most of which we haven't even thought of yet. There's so many different things you can do once you get that really low latency.
4. What is the State of the IoT Market these Days? Is it Still in its Infancy, or Nearing Mainstream Adoption?(4:18)
Beal: "So I think IoT is certainly picking up, if I bifurcate it into IoT for consumer and IoT for business, let's look at them maybe differently for a second.
"Certainly this past holiday season, with security concerns, the reality is that personal voice assistants, doorbell cameras, these things are exploding in popularity. And interestingly enough, with IoT, the 5G providers want to connect those devices directly to 5G. There's this idea now that 5G and edge might eliminate the need for wifi, it might just replace wifi, and now that means the network has to be able to support an IoT device on the same network as maybe a device that's more important.
"So from a consumer standpoint, I think IoT is really exploding. And then there's the business market and IoT is now going into every device on a factory floor, every device in a business, even the military.
"I work actually with our federal customers as well, and you have now military bases that are trying to figure out what they do when a soldier or a spouse comes in with an IoT device. Do we want that on a branch network? So there's a whole world out there of securing IoT, and then there's a whole other world of connecting it. And does wifi exist forever, or do the 5G operators come in and just say, "Hey, bypass the wifi in the house and put it right on our network?" That certainly is, I'll tell you, one of their goals."
5. What are the Benefits of Edge Computing, for Consumers and Businesses? (5:14)
Beal: "I'll start with consumers, and then we'll talk businesses. So I think for a consumer one of the benefits are it's just a better user experience across the devices that you'll have in your home. If we think of a home, we've got everything from our Alexa, which is a relatively... It puts low demand on our network, all the way out to our kids are doing AR and VR gaming or even just any gaming where latency is a concern, and edge compute is going to enable some really neat use cases.
"If I can go back to gaming for a second, if you could get down to sub-1 millisecond or even sub-5 millisecond gaming, that's going to enable an entirely new industry for gaming, and even cloud-based gaming. Google is getting into cloud-based gaming now, Microsoft is getting into cloud-based gaming.
"We might not have that console that we're using for gaming much longer, it might all be done between edge doing the low latency stuff and like we talked about early on, cloud then doing some of the heavier lifting, and we can play these amazing games without needing a $500 and $400 box hooked to our TV. So and edge is going to be the empowerer of that. And even where containers are going with edge, that is what allows that edge device to run a different application based on what that neighborhood or what that edge is the demand for it at the time.
"So I think one of the biggest ones is, it's going to enable a business to buy new services from that, whoever has that edge device. A lot of times it'll be a service provider. It might be your local cable or telco operator and you'll be able to very quickly instantiate security services, you'll be able to quickly set up network slicing.
"Let's take a factory, for example, that has some IoT devices that are unimportant, and maybe some IoT devices that are very important. You might have an Alexa or a connected toaster in the break room, but you also probably have some robotics in that factory. Those are both IoT, but they're very different types of IoT.
"Hospitals, same thing. CT and imaging machines are considered IoT, which is insane. So if you need an MRI, that's considered an IoT device, but you would want that treated very differently on your network than your toaster. Again, I think that network slicing use, the ability to split up workloads based on demand is a great benefit for a business, and edge will deliver that, and then there's the lower latency. Some use cases really benefit even for a business from very low latency and hauling all the way back to a cloud data center is always gonna be more latency than running that use case right there at the edge."
6. What is the Future of Edge Computing?(4:03)
Beal: "I think edge might also be a real enabler to autonomy, whether that's autonomous vehicles, autonomous drones, I think once we get real low latency compute out to the edge, on edge compute, it's going to enable certainly autonomy, it's going to enable better AI and ML, artificial intelligence and machine learning, because again, you're gonna have more compute resources available out.
"I also think it's going to do a great thing for just sensors on power grids and pipelines, just protecting the national infrastructure. There's a lot of talk about being able to add a lot more sensors to our pipeline infrastructure in this country, our rail infrastructure in this country, telecommunications infrastructure, and being able to have edge compute, I think is going to help secure better and another discussion on why that is.
"So, I certainly think autonomy, security, and then the world of IoT and securing the world of IoT, there's going to be a lot of new things that are enabled by edge in those areas, for sure. And then last but not least, 5G. 5G doesn't work without edge. And we all want 5G, we all want faster phones, we all want better experience on our devices, that just doesn't work without edge.
"We've even done some proof of concepts and some studies to try to see what decisions or what intelligence can an edge device do before it needs to phone home up to the big compute up at the cloud, and you start to realize that with some training, edge devices can become pretty self-sufficient and can maybe make some decisions on, let's say, the 80% of the use cases, and then only phone home for the 20% that it can't figure out.
"So that enables a bunch of things. Like when you go to the airport, you're doing facial recognition, for example, to check in the plane, that's an edge use case, right? It's identifying you super quickly. It's not phoning home, it's doing it right there, right at the gate, so that it can identify you and then send you on your way. That's one example of edge doing what could be a very compute-intensive process. Facial recognition could be very intensive, but you can do most of it at the edge and if it can't quite match, okay, let's phone home, let's figure out, is this Bryan or is this someone else?"