We’re currently witnessing the rise of Edge computing as local devices that capture and store data to become more ubiquitous.
In fact, a report by F5 on the State of Application Strategy for 2021 states that there is now an overwhelming interest in the edge, with 76% of respondents already using or planning to use it. This is driven by the need to improve application performance, provide real-time computing, data collection, analytics, and more.
In this round of Cocktails, our guest gives us a glimpse of what’s ahead for edge computing, and we discuss the various use-cases for it, the main drivers for its adoption, and how organizations can potentially improve security when you place your systems closer to the edge.
Kevin Montalbo: Joining us as always from Sydney, Australia is TORO Cloud CEO and Founder, David Brown. Hi, David! How are you doing?
David Brown: Good, Kevin.
KM: All right. And our guest for today is responsible for education and evangelism of application services available across F5’s entire product suite.
Her role includes authorship of technical materials and participation in a number of community-based forums and industry standards organizations, among other efforts. She currently focuses on cloud computing, infrastructure, devops, data center architecture, and security-related topics.
Prior to joining F5, she was an award-winning technology editor at Network Computing Magazine. And most recently, she was named one of Whizlabs’ Top / GlobalCloud / Thought Leaders / and Next Generation Leaders of 2021.
Joining us today for an “edgy” round of cocktails is Lori MacVittie. Hi Lori, welcome to Coding Over Cocktails! How are you?
Lori MacVittie: Hi, thank you for having me! I am doing really well. Especially after that pun. That was awesome. It set the tone for the entire podcast, I think.
KM: Yeah. we’re going to be talking about edge computing today as per the edgy pun. But I’m not equipped enough to explain that to you. That’s why we have Lori over here on the show. So, let’s jump right in. I think some people would be confused between edge computing and a content distribution network, or even simply having regional data centers for different geographic markets. So, can you please tell us what edge computing is and where exactly is the edge of the internet?
LM: Wow. I always like to say that the edge of the internet is right next to the restaurant at the end of the universe. So, “Hitchhiker’s Guide”, right? It’s out there. It really is. It’s somewhat accurate. I mean, the edge – it’s in my closet on the other side of this wall, it could be on my phone, could be on the devices that are operating my fish tank right now. More likely it’s going to be in a retail shopping center or under a cell tower, but it’s going to be closer than a regional data center. And that’s really what our options are right now, right? It’s regional data centers. There are cloud data centers all over many regions that help us distribute globally. And right then you have your centralized data center.
So, you know, the edge is really pushing out to where the users of applications might be. And whether that user is a device or a human being or on a manufacturing floor is another interesting discussion. The difference between CDN and edge is CDN has a focus. And it evolved out of a need to deliver content. That’s why it’s called a Content Distribution Network back when applications were, you know, a lot of JPEGs and some text and boom. That was it, right? Blink tag was the most animated thing you could do and horrible thing you could do by the way, just as well. Nope. And that started to evolve around 10 or so years ago, as people realized there was both the capability and the need for certain types of application services to spread more out across the internet. That’s why we see a lot of security, like DDoS and DNS protection where the CDNs used to be. Because it was a natural extension of what they did. But what we’re moving to with the edge is about application and data distribution. And that requires a different model. And it includes endpoints like devices and things that have never been included in the solution before. So, it’s really changing how we deliver applications and content and data.
DB: That makes it much clearer. And you mentioned a few applications. I’d like to dive more into those applications. So, you mentioned the manufacturing floor, you mentioned retail and Internet of Things type of devices like your fish tank. So, can you dive more into the applications of edge computing and what is actually running on that now? And the expectation of what’s gonna be running on there in the future?
LM: Wow. In the future, a lot more. Right now, not as much, but there are quite a few out there. Especially in certain industries. I attended an oil and gas automation conference. These exist. It was very good. AI, ML, data and the edge were all a significant part of all of these sessions, because if you have someplace remote that you can’t get to, of course there’s automation. But you also need to have more data and you also need to have something that’s more responsive like, can it reboot things remotely and actually take action? So, these kinds of applications are getting pushed closer to the devices and equipment that actually need the information to act on, or to be told to act because a lot of times – these could be in the middle of the ocean – it takes you a few days to get there from right here.
So, sending a person out, being on call for that is probably not your ideal job. Nobody told you that would be necessary. So, those kinds of applications are already in use at the edge. Of course, you see things like home automation, right? A Phillips Hue light bulb can be connected to with my app, right? I can pick up my phone and I can control that light bulb. Now you have the bridges that can actually connect more multiple lights and multiple rooms and start controlling them that way. And it can actually talk to the Phillips Cloud and be able to bring down special kinds of lighting configurations for holidays. So, I don’t have to do it.
DB: That’s an interesting example. A lot of people would be familiar with the Hue lighting system. So, you would class that on the edge computing?
LM: I would class. Yes. I think it would all fit under computing, but particularly the bridge, right? It’s a little beefier. It’s running an application. The question is can I run an app on it or could someone else, or is it very specialized? But it is an application. People do it with Raspberry Pis, right? Arduino, right? All the time, they have these very small amounts of compute that they put a specific app on and then run it in their home. Or at the cabin. Maybe in their cars if they’re really lucky. So, those kinds of things become the edge and that extra compute or the ability to run the app there is very important to the concept of being able to leverage it for other purposes, like business, right? We think, “How do I get to get my app out here?”
Well, you know, that’s probably not going to happen tomorrow. But a future view is if we can run all of these other kinds of applications on these devices, very small confined areas, why can’t we do that with applications that need it? Why does the retail store need to have constant connectivity with corporate headquarters, wherever that is? And why is it when that connection is lost, I can’t get my groceries?I don’t like this situation. Why don’t you have a more local version of whatever application you need? And it can synchronize with corporate at different times, but make sure that everyone local can actually buy their groceries.
DB: Yeah. This concept of synchronization with corporate, so coming back to that analogy of the Hue lighting system, you said we have a bridge device locally, which obviously, you know, it was managing some data in some form of the application of the home automation of the lighting system, but it is communicating with the cloud based system as well. So, there is some sort of isolation between the responsibilities of data or if the edge device and cloud still. Is that right?
LM: Yes. I think that’s a very general statement, but true. What I’ve been kind of focusing on as I can get the time to do it is the different types of edge application patterns that are out there, because there are very distinct patterns already emerging around devices that we use almost all of them take advantage of the cloud. Somehow, whether it’s just to manage configuration or it’s actually controlling my device, I have a variety of devices that each interact with the cloud and themselves in different ways. So, I think there’s a very specific set of patterns that we’ll be able to apply that will depend on what kind of connectivity you expect to have, what kind of data you need and what kind of responsiveness you need with respect to that data. Some things require very quick intervention or notification. Other things, not so much, right? You can wait for that. So, that will all play into which of these edge application patterns you may want to apply for your specific use case.
DB: I’d like to take a couple more of those use cases. We talked about retail and the manufacturing floor. Can you just give us a couple of examples, how they would be used in those environments?
LM: Be used in those environments? In retail?
DB: Let’s focus on the manufacturing floor. You said the manufacturing floor is an interesting use case. So what are you seeing happening there?
LM: There’s a lot of devices and sensors that are monitoring equipment. I don’t know if you’ve ever been on the floor of a foundry of a manufacturing plant that makes, say, toilet paper, because there’s a lot of that in my area. There are the machines that have taken over to actually do most of that work are incredible, right? They do almost everything, but they also need constant supervision. Who knew machines need supervision? There are sensors and monitors that are constantly gathering information, data about the temperature, about the operation, how much oil is in this one, does this need lubrication? How long has it been working? All that data has to go somewhere to basically to the edge. There’s an application, that’s gathering it all, analyzing it and sending out warnings or, “Hey, it’s almost time for maintenance”, right? Whatever.
But it’s also a point of alert. If something happens, it can also turn off a machine which, when you have people and machines mixed together, especially if you’re cutting things like cardboard or paper, there’s a potential for a real harm to be done. So, they have to be able to react and say, “Turn that off now. Stop that. Alert someone”. So, they need to be able to react very quickly. That is not something you want to have disassociated from the actual location. You don’t want that even regional, right? You certainly don’t want it in a data center across the world. It needs to be close because you can’t take the chance that that communication might fail because people’s lives are on the line. The healthcare industry is also very much the same, right? Tele-health is one of the biggest growing smart things and use of these kinds of technologies, but also at the edge because hospitals are local and regional.
And then there’s the big ones. So, you need to be able to have this kind of functionality closer to where you are, especially if you live in the middle of the woods, two hours from a regional hospital. And all you have is a very small one. If they have the ability to do a lot of this processing and manage it locally, then they can respond much faster, make better diagnoses, make decisions faster that allow them to maybe send someone onto a regional hospital or let them stay depending on how fast they can get that answer and whether that impacts the prognosis. So, yeah, I don’t think there’s an industry that won’t find a use-case for the edge. It’s just a matter of right where that is. Right now, it’s very focused on devices because that’s the most obvious, and there’s a whole lot of use of applications and devices that I guess aren’t time-critical because they haven’t had that yet. Because most of the time we see devices and they talk to the cloud. It’s not always the most responsive. It’s not always going to give you good performance and get a response back timely enough. So, the capability to do it at the edge is actually going to drive development of those new use cases because they can trust that it’s going to be fast enough and that it is going to have good conductivity because it’s local.
DB: Speaking of that connectivity and speed, I’m guessing the 5G is driving adoption of edge computing particularly in those public devices, like you say, telemedicine and the like. Obviously on the factory floor, you could probably still rely on Wifi networks and the like. But is that what you’re seeing? Is 5G driving adoption?
LM: I think 5G is enabling development of solutions to those use cases that is driving adoption. So, it is, but it isn’t. It’s one of the enablers and it has to be because one of the other things that has kind of driven that evolution from CDN to service delivery to application delivery is that networks for us have always been asymmetrical. I could always pull more than I could push. So, if we had to take advantage of it and our solutions were based on that being kind of the way things were. 5G says, “No, it’s going to be the same pulling or pushing. Download or upload”, which is very important since we know that the number one, I think traffic source downstream is like YouTube and Netflix and streaming services. And the number one upstream is like Tiktok or Instagram. It’s where I’m taking a video and I’m pushing it. These are very chunky pieces of content that we’re throwing around. So, you really need symmetry in your bandwidth so that nothing gets lost basically or pushed out on the way down or up. So, 5G is going to make sure that happens, but also that all the other things that need that network can use it. So, hopefully, you know, if you don’t have four teenagers in your house, I think 5G is going to work for you. If you have more kids, sorry, your bandwidth is going to be gone. That’s it.
DB: Let’s now talk about F5’s acquisition of Volterra if we can. So, can you tell us this was a major acquisition? Can you tell us the reason behind this acquisition and what you’re looking forward to from that partner?
LM: Okay. See, because we didn’t have a V we had an F and an N and an S. We bought a V, we bought a V like on the Wheel of Fortune. So Volterra actually solves multiple challenges that we see in the market that we wanted to accelerate. We know that managing applications and multiple environments across more than one cloud has been a challenge for people. We do a survey every year and they told us “This is hard. We don’t like it.” It’s a challenge for us because every cloud has its own API, its own console, its own model and way that you manage it. So, if you have more than one, you’ve got a lot of complexity trying to figure out “How do I make this one thing work the same in two different environments using two different systems?” It’s very difficult.
So, Volterra, by virtue of its ability to virtualize public clouds and private clouds, its own data centers, basically says “You all look the same to me” so I can deploy on all of the clouds if I want. And I’m assured that it’s all gonna look the same. I know how to do it. No matter what cloud it is. You can start a cloud tomorrow and if Volterra supported it I could use it. That would be awesome. So it solved a multicloud challenge, but it’s also edge. So, we see that future of edge really growing, that it’s going to continue to expand. 5G continues to broaden and get more adoption. It’s going to result in even more applications and users and things.
And we wanted to make sure that we had a way to actually build solutions that took into consideration all of the things about application delivery that F5 knows really well. Security optimization, how to scale it, those kinds of things that F5 has done really well for years, we might be able to extend that to the edge, wherever applications may go. So, Volterra lets us do that and solve the multicloud challenge at the same time. So, we like where it’s going and we’re hoping to be able to continue to do those kinds of services atop it and be able to respond more quickly to other needs and other services that people might find they need as edge starts to evolve because there will be new services. No one’s thought of yet where we say, “Wow, we need that. So let’s go build stuff.”
DB: Yeah because there are a number of challenges, right? So, you know, part of cloud adoption, the attractiveness of cloud adoption is you’re pushing responsibility of the management of your servers out into the cloud. It sounds a lot like when you’re bringing devices and possibly servers, even if they’re small servers, back to the edge, now I have more responsibility managing all these devices. I have security issues, where data is residing and privacy issues. You recently published a State of the Application Strategy for 2021. You found out how organizations are facing difficulty managing the security of their applications in today’s multicloud environments. As cloud allows data to be closer to users, can it improve the way organizations are implementing security?
LM: Wow. That seems like a loaded question and it really can. I’m going to reach way back into my memory too, like 2009, when we started having discussions about cloud and security and Christopher Hoff. If you know him, he’s really brilliant in terms of security. But someone asked the same question about cloud. “Is cloud really secure or can cloud really help?” And it comes back, I think with edge, to the same answer he gave, which is if your security practices are good now, then as you go to the edge, they’re going to be good. If they’re not so good, well, you can see they’re not going to be so good no matter where you are. So, a big chunk of this depends on security practices and no matter how good the solution is, you can’t force someone to adopt those security practices.
As we’ve seen so many cloud misconfigurations of S3 buckets have caused so many breaches. That’s not a problem with cloud or cloud security. It’s a problem with people and process and practices. So, could it add improved security? Yes. We’ll have to wait and see. We also don’t know yet necessarily what new challenges edge will bring. There’s a lot of chatter about being able to do more P2P. So, more peer-to-peer applications and uses on the edge, if we’re all on the edge and we can all run apps, we can all be peer-to-peer or sharing.
DB: You raised some very valid points there. You’re saying it could improve security. How?
LM: Okay. How? So, I think, because at least from our perspective, one of the things that we want to make sure as we move forward is that a lot of the security is in the platform. So, it’s not a bolt on, it’s not an add-on. It’s there and it always moves with the application so that no matter where it goes, it’s getting the same security in the same protections, whether that’s across clouds or just from one edge location to another. So, building that kind of security in or enabling that will help improve security in the sense that it’s there. And it’s easy to put in. We all know that security. Nobody likes it because it’s hard. “It blocks me”, “It slows me down”, “There’s a gate”, “Somebody wants a test”, “I don’t like this. I’m going home”, “I still haven’t finished my latest secure development lifecycle training. I’m sorry.”
And that’s true and I am sorry, but we do. We put that off and we do that in the cloud and we’ll probably do it in the edge. So, if we build it in and make it easier to use, then I think we’ll see better adoption of it, which will ultimately lead to more security or more secure applications and conversations as they move around. Second, you mentioned the privacy of data, right? That is a big deal. Right now, that prevents a lot of movement of data out of certain regions in the world. And we’re going to see that at the edge that will get even trickier, but sometimes having to pay more attention to the data and where it is and who has access to it will result in better security naturally because you’ll pay more attention to it.
Data is something we’ve never really paid attention to as an industry. It’s always been in that database somewhere and it was locked behind a firewall inside and mainframe, it was safe. Now it’s coming out in the world and we’re going to have to start paying attention to it. And that should draw a little more attention to the security of it and how we protect it because it’s no longer just our customer data and transactions. I mean, that’s my personal, private data, my health data, my kids’ pictures. This stuff is important. I don’t want that out in the wild.
DB: So, as I understand it, that private data can stay on the edge. And if we made analytics and the like in the cloud, what we can do is just send back the anonymized data or the aggregated data, as opposed to sending back the private data and then anonymizing it or aggregating it.
LM: That would make it safer. Yes. Right. If it was anonymized first, distilled down. One of the edge patterns is really focused on just that kind of use case, whether you’re analyzing and distilling it so that you’re not saving as much. If I’m a sensor and I’m generating status data every second, do you really need to store that all in the cloud? Does somebody have a calculator? How much is that going to cost? And just for me, that’s not feasible. It doesn’t make sense for the provider. And it doesn’t make sense for me.
But if you have something at the edge that is looking at that status data to make decisions and then distilling it down, anonymizing it, aggregating it, doing some of the processing at the edge, it can then send that back maybe once an hour or even once a day. So, it’s going to be more cost effective. And it’s also going to be more secure because you’re not constantly sending something back and forth and sending everything back and forth. So, I think you have a good point. That’s one of the ways that will make our data a little more secure because we’ll do more processing of it before it ever gets to wherever it’s ultimately going.
DB: F5 is talking a lot about edge 2.0. So, from my reading on F5, they talk about version 1 was basically CDN networks and the evolution of CDN. 1.5 is where it became a bit smarter with applications. And now, F5 is promoting this concept of edge 2.0 and one of the concepts inside there is the unified control plane. So, I’d like to talk a bit more about that because this seems to me, with all of these devices on the perimeter, you’re going to have this huge management headache. Is that what we’re talking about with edge 2.0 with a unified control?
LM: At the moment, what we’re talking about is across clouds and environments, right? Any cloud, any edge, wherever we might be able to virtualize a resource and make it part of what’s really more of an intelligent mash than any kind of specific environment. So, the control plane across that, that’s part of what would enable you to be able to just deploy anywhere and not have to worry about it. It’s one control plane for all environments, but then you get into the application space. How are you going to do that? And it’s got to extend to that, to the entire application management and all of the different services that are included in it.
You have to be able to have one control plane and a consistent set of right policies in order to improve your security and actually control things in a way that makes sense because you’re right, we’re not only distributing applications, but we’re also talking about decomposing them in a lot of ways, which is great to create more things. You have to manage more points of potential failure and also more points of potential attack. So, unifying the control plane is going to provide people who are actually using edge, better control, more consistent control. And it makes it easier also for us to be able to provide the kind of services that are needed to protect against all those things, because it’s unified and it’s always the same.
DB: And this unified control plane, is that through part of the acquisition of Volterra?
LM: I think Volterra is the start of it because it already does the multicloud and allows you to do that. It has that abstraction layer. We don’t have that integrated to everything F5 does yet, let alone enabling any ecosystem, which ultimately I think you have to do is enable others because F5 is great and everybody wants to use it, but we don’t do everything. So, you’re going to need something else. That’s going to have to be opened up eventually, so people can actually attach bolt-on, offer new services as well. So multicloud right now, but not everything yet.
DB: I understand edge 2.0. What else does it include in the concepts there?
LM: Wow. We kind of touched on some of that with the security being embedded, but also that kind of goes along with analytics being embedded in the platform, right? You shouldn’t have to be deploying agents to do monitoring for all of these things when you’re deploying somewhere. That should be just a part of the platform that you deployed your application on. It should be able to report back any of this information that you need about how it’s performing, how the load is, what’s going on, is there a failure, is anything wrong?
So, edge 2.0 will include embedded analytics that are able to aggregate all that data and be able provide you information that ultimately will enable better placement of workloads. So, you’ll be able to start saying, “Wow, it costs too much on this cloud. I’m going to run it on that cloud: or “It’s performing really poorly today here, move it over there” or “We suddenly had a sale on Axe in New York. So make sure we have extra application presence in New York. And then if we have to, move it to California” or move it across the globe, maybe somewhere in Hong Kong. It’s more of that intelligent workload being able to dynamically move it based on the data that’s coming back from the platform. That’s going to be a part of it. So, you don’t have to do that separately or build a different solution. I’m trying to think through our cool list.
DB: There is so much to edge computing and we’re still in the infancy here. A massive growth area. Lori, you have a real knack for distilling something quite complex into very simple, easy to understand terms. I can see you’ve been at F5 for a number of years, right?
LM: Yeah, 14 in real life, which is like a thousand and technology years.
DB: There’s not many people that could say they’ve been in the same company for 14 years now. As an evangelist of F5’s products and services and the technology underlying and what you’re doing there, you’re very talented and good at explaining it. Thank you so much for your time on our podcast today. Where can our listeners follow you and learn more about what you’re writing and talking about?
LM: Well, first thank you for inviting me and letting me talk for however long we’ve been talking. I actually don’t like to talk, so it’s really weird that I do like to talk when I’m doing this. So, thank you for letting me do that. People can find me on Twitter at “lmacvittie”. Same on LinkedIn. F5 does have a blog, but it’s a long URL so I hate to like, you know, try and spell it out, but you can find them. I do share them on LinkedIn and Twitter. And of course the F5 corporate social media will always share that content out.
DB: Good. Thank you very much, Lori.