Today’s episode features an interview between your host Matt Trifiro, and David Xie and Dalerie Wu of Zenlayer. In this interview, David and Dalerie discuss the unique challenges of bringing edge solutions to emerging markets, and how the edge represents a huge opportunity to introduce a quality digital experience to countries with underdeveloped telecoms infrastructure.
Today’s episode features an interview between your host Matt Trifiro, and David Xie and Dalerie Wu of Zenlayer.
David is the Chief Product Officer at Zenlayer. Before joining Zenlayer, he spent eight years at Gartner, most recently as Group Vice President of Digital Products.
Dalerie is the Head of Corporate Strategy at Zenlayer. She has been with Zenlayer since 2015 and previously led its global marketing team.
In this interview, David and Dalerie discuss the unique challenges of bringing edge solutions to emerging markets, and how the edge represents a huge opportunity to introduce a quality digital experience to countries with underdeveloped telecoms infrastructure.
Key Quotes:
“I'm sure a lot of technologists think about the edge from the technology angle. We tend to think about edge purely from our customer's perspective. Because we are serving global companies, they are usually doing well in their home country, which is usually in the developed world. For example, if the United States is the core market, then the emerging markets will become the edge for them.” -David
“All of those emerging markets are growing at a much faster speed and they have tons of people, so if you add those two things together and then couple that with the poor infrastructure, that means there's a huge opportunity for infrastructure providers to bring a good or decent digital experience to those users.” -David
“In emerging markets, it's really people using mobile internet more than wired connections because of the infrastructure issue. It's hard for them to build fiber, so you have wireless internet being so much more dominant. Over 90% of people who are on the internet use a mobile device to access it.” -Dalerie
“I think once [5G] does roll out to emerging markets, it will have a much bigger impact than it would probably have here in the US…[but] really building and ramping that up is going to be a challenge.” -Dalerie
“Cloud and edge [are] complementary. Cloud is always going to be here as the core, but as more and more applications are being consumed by users all over the world at the edge, you need the edge computing component to complement that.” -Dalerie
Sponsors
Over the Edge is brought to you by the generous sponsorship of Catchpoint, NetFoundry, Ori Industries, Packet, Seagate, Vapor IO, and Zenlayer.
The featured sponsor of this episode of Over the Edge is Zenlayer. Improving user experience doesn't have to be complicated or expensive. Zenlayer helps you lower latency with on-demand edge services in over 180 PoPs around the world. Find out how you can improve your users' experience today at zenlayer.com/edge
Links
[00:00:00] Matt: [00:00:00] Hi, this is Matt Trifiro, CMO of edge infrastructure company, vapor IO, and co-chair of the Linux foundation, state of the edge project. Today. I'm here with David Xie and Dalerie Wu. David is the chief product officer of Zenlayer and Dalerie is the head of corporate strategy at Zenlayer. We're going to talk about their backgrounds in technology, Zenlayer's plans to accelerate the edge in gaming and media and the future of edge computing in emerging markets around the world.
[00:00:23] Hi David, and Dalerie. Thank you both. How are you guys doing? Doing great.
[00:00:27] David: [00:00:27] Thanks a lot for having us.
[00:00:28] Matt: [00:00:28] Yeah, this was really fun to do a three-way interview. So let's start with Dalerie, Dalerie how did, how did you get into technology to start
[00:00:35] Dalerie: [00:00:35] with? So I've been interested in tech, I guess, ever since I was in high school, I actually took a computer repair class in high school.
[00:00:42] I was the only girl in the class until I drag my best friend to join with me. So that's how I got started. I also, actually, I started off as a computer science major in college. It didn't last very long because I found just coding to be a little bit dry for me. So I did [00:01:00] change my major to industrial engineering, but I've always been interested in
[00:01:03] Matt: [00:01:03] tech.
[00:01:04] That's awesome. And how about you, David? What was your path to technology?
[00:01:07] David: [00:01:07] Yeah, so it's a really after college, I worked as a developer for a very short period of time, but in my interest has always been. On the problem side rather than the solution or the technology side. So I'm, uh, a lot more passionate about solving problems using technology rather than the technology itself.
[00:01:27] And
[00:01:27] Matt: [00:01:27] an interesting part of your biography is you came to Zenlayer via Gartner. So you actually have seen both sides of the equation.
[00:01:35] David: [00:01:35] It's absolutely true. Actually, before Gartner I worked on the technology side for about seven years. As an architect that is with a company called Openwave. So this amazing company that opened ways to pioneer of mobile internet.
[00:01:51] So we try to solve a pretty intriguing problem of enabling internet browsing over a very slow network, which is kind [00:02:00] of similar to what a den they are trying to solve. But in that back in the late nineties, Of course, like many people are hoping to get on lines for their phones, but at the same time, you, you all know that the phones can only do three things at a time.
[00:02:14] One is making phone calls. The second is to receive and then send short messages. And then the third is to play that a snake game. But the network back then was just like so slow. Even the two G network, which is the best editor at a time. It's at a speed of about like 40, 50 kilo bits per second, which is like way too slow for browser.
[00:02:40] My company kind of created a product called lap. Right? You remember that? Oh yeah. Yeah. I remember burlap almost all domain names. You have like a www version as well as version. So that was a fun job. Uh, so I helped a lot of welders carriers enabled mobile internet traveled around the world. So tons of fun.
[00:03:00] [00:03:00] But then after seven years working in the field, the technology field, I realized that at least the excitement to me comes from seeing people getting online through their phones rather than the technology itself. So then in reflection, I know that I probably shouldn't be working on building things myself, but rather focusing on figuring out what problems to solve and then rely on my team to build a solution.
[00:03:28] Matt: [00:03:28] Well, that's great. I mean, salary, you know, Zen layer, one of the ways I think of it is maybe not bringing more people online, although I'm sure that happens too. I think of Zen layer as, you know, bringing the rich internet to more people. Can you tell me a little bit about what San layer does? I mean, you've been there since 2015.
[00:03:43] Dalerie: [00:03:43] Yeah. So the core of what we do is improving user experience. So like you said, I mean, we don't build the basic infrastructure that bring people online, but we want to make sure that whoever is online can experience the internet, can experience the applications [00:04:00] as well as people, anywhere in the world. So we focused on emerging market users where infrastructure is typically a little bit less developed.
[00:04:08] So how can we ensure that those users also can experience the internet as well as the developed countries? So let's,
[00:04:14] David: [00:04:14] let's,
[00:04:14] Matt: [00:04:14] um, let's take a specific, you know, before the call, we were talking about Indonesia and maybe you could explain, you know, the sort of internet evolution of Indonesia and how Zen layer has helped, uh, improve the user experience there.
[00:04:28] Dalerie: [00:04:28] Yeah. So I'm actually Southeast Asia in general, it's growing very fast in terms of internet adoption. It's the fastest growing internet economy in the world. The internet users is expected to double by 20, 25. And so we're looking at just a massive market. And so we have a lot of companies that are interested in tapping into Southeast Asia and Indonesia is the largest market there.
[00:04:51] It has the most population and just. A lot of users, all of young users that are very interested in getting in to the internet to [00:05:00] videos, to gaming the entire infrastructure in Indonesia is actually pretty fragmented. So there are a few major carriers, but they don't really interconnect. With each other due to administrative or political reasons.
[00:05:13] And so just in Jakarta, actually we have eight data centers in Jakarta and we are directly connected to multiple carriers in the city and we've built our own backbone to a fiber infrastructure. I feel like to connect all of these carriers and so that we have this interconnected web. And, um, that was our first foray into Indonesia was, um, to Jakarta.
[00:05:36] And as we built out our infrastructure, because Indonesia is actually composed of 17,000 islands. Wow. Yeah, it's the largest Island country in the world. And so mind
[00:05:47] Matt: [00:05:47] boggling. It
[00:05:48] Dalerie: [00:05:48] is, it is. And so it's hard for the infrastructure there to cover all of the islands. So you really need to be in different areas of the country and what are to try to cover the entire country.
[00:05:58] So the biggest two [00:06:00] islands are Sumatra and Java. We are on both islands. We have. Three pops now in Indonesia, one in Dunn, that's in the West is very kind of Western corner. And then we have, of course our, um, key Pop's in Jakarta and then to the East, we are also in Serbia. And so that sort of covers most of the Indonesian users.
[00:06:20] Pretty.
[00:06:21] Matt: [00:06:21] Yeah. That's interesting. And so when you say eight data center locations in just Jakarta, I think is what I heard. What are you putting those data centers? What's actually in
[00:06:29] Dalerie: [00:06:29] there, we have servers, we have routers, which is basically we have equipment that helps companies to connect to, uh, the local carriers and so that they can send their content out to the local
[00:06:41] Matt: [00:06:41] eyeballs as a customer.
[00:06:43] And maybe David, since chief product officer, this is a great question for you as a customer. Like, what are your products look like? I mean, what am I, what services or products can I purchase from you and who are your typical customers? Yeah.
[00:06:53] David: [00:06:53] So maybe even taking a step back and then think about a, uh, let's take a gaming company as an [00:07:00] example.
[00:07:00] We actually serve quite a, feel, large gaming companies around the world. Usually what's happening is that they are doing pretty well in their home market. But in, as you can imagine, very soon, the home market will become saturated. And then they're all looking for high growth rate. And then know of course that the emerging market becomes a pretty good target for those gaming companies too.
[00:07:24] Reach out to those, um, emerging market where the infrastructure was bad. You're really two fundamental different approaches that those companies take. One is they deploy their computing capabilities to the local market so that the latency can be short enough to sustain a pretty good gaming experience.
[00:07:44] So that we have those data centers that allow the gaming companies to deploy their computing capability. And will they
[00:07:51] Matt: [00:07:51] deploy their own servers typically?
[00:07:53] David: [00:07:53] No, they don't. Usually the reason is the lead time to deploy. They are local [00:08:00] to those local data centers will be, you know, a couple of months. And then what's interesting about the gaming company is that the gaming industry is that it's sort of a high investment.
[00:08:11] High return and then high risk business. So you can develop like 20 different games throughout the year, but in only one make to the top 10. And then once they become a top 10 game explodes, it's both. Uh, so you will like all the sudden ask for like 2000 servers in the data center grow from like 200 to 2000.
[00:08:33] And then if you wait for a few months, then someone, then someone else it's going to take you a year. How do you solve that problem? Essentially, what you would need to do is to go with a vendor infrastructure vendor that has the large scale so that they have to vet the servers, the spare servers in the data center, that's ready for a gaming company to pick up.
[00:08:53] That is the first approach. The second approach, like some gaming companies, they, uh, they decide not to, um, [00:09:00] Deploy to the local market, but then they figured out a way to make the connection, the connection between the emerging market and then their home market a lot faster so that even though they have their computing devices at home, the users in the other game, players in the emerging market can still have a sustainable, reasonable gaming experience.
[00:09:22] By leveraging the fast network. That is the Otter service that we offer. So
[00:09:26] Matt: [00:09:26] you offer the bare metal compute if I want it. So you're, it could be an infrastructure partner, a server infrastructure partner for me, and help me scale, you know, in days instead of months. But you also can provide me the optimized network routes, which I imagine even if I'm deploying in the market, I still may want, of
[00:09:41] David: [00:09:41] course.
[00:09:41] Yeah. Like even if you deploy through a BMC, a bare metal cloud you still need sort of the backbone to connect to your home computing clusters. Yeah.
[00:09:51] Matt: [00:09:51] So gaming is a great example. Is there another emblematic industry that Zen layer tends to serve really well into these emerging markets?
[00:09:58] David: [00:09:58] I can start with, uh, [00:10:00] another example and then Dalerie has been with the company for much longer.
[00:10:03] So I'm sure like she will, media is another kind of example you think about Netflix. Think about Tiktok and then the whole array of media companies trying to basically. Allow worldwide users to share and then enjoy the content to get that massive amount of amount of content distributed to the end users that they will require tons of infrastructure support.
[00:10:30] Matt: [00:10:30] So are you also running what looks like a CDN or a caching service? That is
[00:10:34] David: [00:10:34] absolutely right. Got it. CDN is one side of the equation. But in a, what, what people may not always know is that there's also the dynamic content, so that it is more like real time. So for that, then a CDN, like typical CDN technology will not work well.
[00:10:51] So you need to be able to. Accelerate the dynamic content delivery
[00:10:57] Matt: [00:10:57] as well. Yeah. Want some bare metal [00:11:00] servers to do the dynamic work and have some caching servers to do the storage work and do that in a local market. And what you really do uniquely, it sounds like, is focused on these emerging markets that, where there's this high growth, but also this lacking infrastructure.
[00:11:12] And you go in and sort of patch the pieces together like Dahlia was saying, you know, stitch together, these networks that aren't talking together to each other yet dally David suggested you might have a third example.
[00:11:22] Dalerie: [00:11:22] Yeah, well, I actually wanted to add just a little bit. I think companies that really would benefit from what we do is companies that really care about real-time interactive applications.
[00:11:32] And so other than gaming, we're looking at media entertainment, so live streaming, but also online education. When COVID hit, everybody just did remote learning. And so there were a lot of students actually in Tyna horse or stranded there, they couldn't come back to us universities, but they still want to take classes.
[00:11:51] Their classes were still taking place online. So, um, what we do is, you know, we help that as well. So it's really the interactiveness of your [00:12:00] applications that would really benefit from
[00:12:01] Matt: [00:12:01] what we do. You know, what's interesting is, you know, I've been in this the edge industry for as long as it's been around, at least in this wave of it.
[00:12:08] And, you know, we've stopped asking the question, define edge computing. Cause I think we've all realized that there are many edges and it really depends on your application and your infrastructure and a bunch of other things. But one of the, the metaphors that I've found really powerful, especially when talking about.
[00:12:24] The infrastructure side of the equation, the infrastructure side of the last mile network, there's the last mile networks and the whole set of edge computing that happens down, you know, in the field or on the oil rig or on the cruise ship or all that thing. There's a whole bunch of infrastructure, side, edge computing.
[00:12:36] And the way that I started talking about it is this kind of conceptual internet backbone, right. This like really rich. Interconnection that started in two places may West and may East, and has grown into these other places. And it kind of ends at least like in the U S for example, it ends at about eight major cities, like Los Angeles, Miami, New York, Dallas, Denver, and Boston, if you're generous.
[00:12:58] Right. And then [00:13:00] you've got sort of, you know, legacy infrastructure that doesn't look like a backbone doesn't have core routers. Doesn't have rich interconnection. But when you look outside the United States, you look into these, you know, emerging markets and the edge, you know, might not even have made it to Jakarta, let alone some of the other islands.
[00:13:17] And so in some ways what you're doing is going, you're extending the internet backbone into those markets and your edge today is. In those emerging markets. And I imagine within those emerging markets, you push your edge farther out. Like you said, in Indonesia, you know, you started in Jakarta and you went to these other, other things.
[00:13:33] So when you think about edge computing, is that a way that you think about it?
[00:13:37] Dalerie: [00:13:37] Definitely. I think we start off with kind of just one location, one pop in a country, but as that country gains more users and as our customers are demanding better user experience for their users in those countries, we have to spread out into other cities and other locations.
[00:13:53] And so the edge kind of gets pushed. Further and further to, uh, gain closer to the users. And another example [00:14:00] that I can bring in is that in India. And so Mumbai is actually the city where most submarine cables end, and that's the most interconnected city in India. And that was also our first pop there.
[00:14:10] And so we started India about six years ago. Chennai is a second distant second to Mumbai. So tonight, you know, it's on the East coast of India. Mumbai is on the West coast. And so we then expanded into tonight and tonight really connects to more as an Asia pack countries from India, whereas Mumbai connects more to the middle Eastern Europe.
[00:14:31] And so now we have two locations where we're sort of connecting India with the rest of the world, but also as we delve deeper into the Indian market, we have to now have 11 pops. I'm not going to attend a name, all of where they are, but they're all over India. So it's North, South, East, West, and, uh, last year she late last year, whenever customers wanted to broadcast a live stream, a cricket competition in India.
[00:14:56] And so, um, to the India users. And so they used all of [00:15:00] our pops. We spend things up for them really quickly, just over, you know, a couple of days. And they use the service for maybe a few weeks and then D wrapped everything down. And so it was a festival thing that we can provide to the customer when they
[00:15:12] David: [00:15:12] needed it.
[00:15:13] Yeah. Just to quickly add on to what Dalerie mentioned. We think about edge as a company, we. Tend to think about ad purely from our customer's perspective? Well, I I'm sure like a lot of technologists think about the ad as like from the technology angle, in our case, because we are solving global companies, they are usually doing well in their home country, which is usually in a very.
[00:15:37] Developed world. We tend to think that the whole, for example, the United States is the core market for them. And then the emerging market will become the edge for them for all kinds of good reasons. Infrastructure is certainly one big, big barrier for people to get into those worlds. And if you look at the cross country connection, bandwidth of those [00:16:00] emerging markets, They're very crappy.
[00:16:03] Take India as an example, like a per person, average, we're talking about like seven megabits, but then just as a benchmark for us is about a hundred K for UK is 500 K. So those emerging market, they have much crappier infrastructure. So you want to get into those, then that will become your edge for our customers.
[00:16:24] Matt: [00:16:24] Yeah. I mean, that makes a lot of sense to me, based on your observations, what are some of the trends in emerging countries that are most interesting to my listeners, but also to your customers?
[00:16:34] David: [00:16:34] The reason that we call those emergent market is really for two, well, maybe three, one, they are fast growing, you know, China, the economy is growing at about, uh, you know, it has been double digit for the past 30 years.
[00:16:47] So down a little bit due to COVID, but this year it's going to be at 8%. India, 12% Indonesia, 6%. So all of those emerging markets are growing at a much faster speed. [00:17:00] Second is they have tons of people, China, 1.4 billion. So it's India, Indonesia. We're talking about like close to 300 million people. So if you kind of add those two things together, of course, it's got a lot of market potentials.
[00:17:14] And then coupled that with the poor infrastructure, that means there's a huge opportunity for infrastructure providers like us. To bring a good or decent digital experience to those users so that they can, um, they can become a big market for players at home. Yeah.
[00:17:33] Dalerie: [00:17:33] And to add onto what David said, in addition to just their internet growth is a lot faster than Bella markets.
[00:17:41] They may have a lot more people, but also they spend a lot more time on average on the internet. And so, um, We are social and who sweets, they put together a really interesting report every year called digital, um, you know, 2021 is this year's title report last year was digital 2020. And so there's really interesting [00:18:00] statistic in that report that said, um, in Latin America, for instance, on average people spend nine to 10 hours a day on the internet.
[00:18:09] Whereas in the U S it's more like six to seven hours. And so you have some of these users in emerging markets that are spending half their day on the internet. So you have a lot more of their attention span on the internet as well.
[00:18:21] Matt: [00:18:21] I think I'm on the internet nine to 10 hours a day to
[00:18:28] put a server in my bedroom. I got my own little emerging
[00:18:30] Dalerie: [00:18:30] market here. Matt, do you want to guess, um, what is the country where users spend the most time on the internet?
[00:18:35] Matt: [00:18:35] Oh, that's interesting. I wouldn't even know where to start. I wouldn't even know where to start. You got it. You don't have to tell me. I'm not even gonna guess.
[00:18:42] Dalerie: [00:18:42] So the Philippines that's a good reason. I don't know why, but they spend more than 10 hours a day on the internet on average. So users between ages 16 to
[00:18:51] Matt: [00:18:51] 64 may, maybe an age thing because my, my kids spend, if I didn't kick them off, they would basically their entire waking life on the internet.
[00:19:00] [00:19:00] Dalerie: [00:19:00] Yeah. And actually, that's another interesting thing I wanted to mention is that.
[00:19:04] In emerging markets, it's really people using mobile internet more than wired connections because of the infrastructure issue. Right? So they, it's hard for them to build a fiber. And so you have just a wireless internet being so much more dominant and over 90% of the people who are on there to actually uses a mobile device to
[00:19:22] Matt: [00:19:22] access it.
[00:19:23] And that's one of the most interesting things, because you hear a lot about 5g when it comes to edge computing and edge computing is in 5g or tightly in Iraq because 5g is actually. And edge computing application, right? Because the virtualized network functions, you're running it on servers. Those servers have to be out at the edge in order to run the virtualized network functions.
[00:19:44] But it also is an enabler for low latency last mile connectivity. And so I could see 5g. Having a, in many ways, a much bigger social impact in emerging markets than it will in, you know, whatever existing [00:20:00] markets or, you know, the sort of the tier one markets like the United States and Europe. Do you have any thoughts on 5g and its rollout into emerging markets?
[00:20:07] So I
[00:20:08] Dalerie: [00:20:08] completely agree. I think once it does relative to emerging markets, it will have a much bigger impact than it would probably have here in the U S one hurdle that emerging countries need to get over is that because when we go from two G to 3g, to 4g, to 5g, the range of the wireless network actually gets shorter.
[00:20:25] And so you need a lot more towers and a lot more infrastructure. And so really building up and ramping that up. It's going to be a challenge in some of these emerging markets.
[00:20:33] Matt: [00:20:33] Yeah. And you've got to have fiber to those radios and yeah, there's a whole inch. That's interesting. That's an interesting thing.
[00:20:38] I have to pay more attention to that. So when you, when you look at, from the perspective of Zen layer, like what are the most exciting trends that you're seeing that are sort of driving edge cutie beyond the growth in the emerging markets? Like, are there any other trends that are exciting to you? So,
[00:20:52] Dalerie: [00:20:52] I mean, I, I do think online education is going to be a pretty big driver for us.
[00:20:57] And I say in China, there [00:21:00] are. Two or three online education. Startups are getting huge, massive valuations, and they're planning to go public and online education is really one area where you need such a solid internet connection and low latency and low jitter and, and low lag, right? Because of a t-shirt talks and the student can't hear, or there's a lag or what they're seeing, like their mouth doesn't match the sound.
[00:21:23] It's a very poor learning experience and you want to keep the students engaged.
[00:21:28] Matt: [00:21:28] Yeah, I can see why you're excited about that. I mean, it really has the opera. I'm only education has the opportunity to completely transform areas of the world that don't have access to higher education, or even just basic core education.
[00:21:42] And. Um, and just, just imagine what the world would be like, if we educate another 50% of the population and they're doing, you know, innovative science and innovative social programs and things like that, that's a, that's a really neat vision. And I know that, uh, you know, say interviewed Joe, your CEO [00:22:00] on an earlier episode.
[00:22:02] And, uh, he's passionate about the social aspects of computing. And I thought that was kind of a neat part of your company. Tell me a little bit about the, kind of the, the, the passion of the mission. What is the mission? What is the passion? Our mission
[00:22:15] Dalerie: [00:22:15] is to improve user experience for every person in every organization in the world.
[00:22:20] That's what we focus a lot on emerging countries, because for the developed countries, you have a pretty good user experience already. I mean, if you're in LA or New York, you know, you don't have a problem with your internet connection. But when you are in Africa, for instance, it's a very difficult country to get good internet in.
[00:22:38] So first of all, there's very few fiber infrastructure in any of the countries in Africa. A lot of them sort of rely on satellite connections, which are pretty slow. Um, a lot of them don't really have really any telecom infrastructure. And so they're building a lot right now, but if we can bring, um, Better internet to African users, even within their existing infrastructure.
[00:22:59] I think [00:23:00] that would be a big win for
[00:23:01] Matt: [00:23:01] everybody. Yeah, that's really neat. So do you have any predictions, like if you look out the next 12 to 18 months, any sort of actually, I should ask David because he was in the business of making predictions, what's in the upper right quadrant. In the next 18 months, David, just
[00:23:15] David: [00:23:15] maybe I can take a, like a narrow scope, uh, to answer the question on the sense of, uh, then later, like Joe likes to talk about our vision to, um, quantify the vision, to bring the latency down to 10 milliseconds.
[00:23:32] That is pretty, not very aspirational, but then I did some kind of math around that to figure out like, what does it really take? To bring down the latency to 10 seconds. And at the same time cover billions of people because of course like fixing the latency problem for a small population is much easier than a, I mean the whole world of people.
[00:23:53] If the thing about that, the limitation to the latency, I guess the ceiling, first of all, it's about. The [00:24:00] speed of light or speed of, uh, electronics. Right? So we, we know that the travels at about 300,000 kilometers per second. So what does that mean? 10 milliseconds means that roughly 3000 kilometers. So your data center will have to be next to your user with things 300 kilo meter.
[00:24:20] Of distance having a
[00:24:22] Matt: [00:24:22] fiver
[00:24:25] David: [00:24:25] that we know that there are hops in between like a suspicious routers, they were all adding latency. But so in reality, we are really talking about like 500 kilometers of distance. If you go beyond that, then there's no way that you can bring the latency down to 10 minutes seconds.
[00:24:42] So that means like tons of data centers around the world in order to do anything kind of close to that.
[00:24:48] Matt: [00:24:48] Yeah, that's really interesting. And if you, you know, you look at these and we've done a lot of this research to stay the edge, but you look at the, the kinds of applications that emerge when you get predictable latencies below a certain [00:25:00] amount.
[00:25:00] You know, if you're going country to country on a cloud over the. Traditional internet, not like a private backbone, 150 milliseconds, maybe way more than that. Yeah. Yeah. And certainly not predictable. Right. And that's, you know, I I'm fond of saying like, that's the internet we built and that's fine. If it's mostly humans talking to machines.
[00:25:19] Because we operate ones of seconds. Although do you point it out that in emerging markets, it's not fine for humans talking to machines until you've placed infrastructure there, but we're moving to a world where it's going to be primarily machines talking to machines. And I mean, even 10 milliseconds is glacial for a machine, right?
[00:25:34] So, but there's new classes of applications that emerge as you get, as you lower these latencies. And I think that's going to be one of the most interesting things as the average oral app, maybe the predictable latency in a certain country goes below 30 milliseconds. You're going to see a wave of applications that need that when it goes below 10 milliseconds away with applications that need that.
[00:25:55] And then eventually, you know, I mean, I've, we've got customers that are demanding 75 [00:26:00] microseconds of latency, but for building out a 5g network. So, you know, that's the kind of the, that's a really. Specialized use case for edge computing. So you didn't give me the punchline though. You did the math, what is it going to take?
[00:26:14] Can you quantify it? I mean,
[00:26:17] David: [00:26:17] so I, I would, I would say that, uh, roughly speaking, we're talking about thousands of data centers around the world, the reason that is a it's a lot more than the, sort of the theoretical, uh, besides the. Latency added by routers and switches and like all the kind of hops. We also have talking about cross country connections, which is like pretty horrible.
[00:26:38] Let's see, you kind of mentioned, but when you ping a server in Indonesia from the United States, it's actually way higher than 115 minutes seconds. We're talking about constantly above 200. But then if you try it during the night, when many people get on the internet, try to do things with easily talking about like a few seconds of delay, you know, that may happen as well.
[00:26:59] So the whole [00:27:00] world is not like, right. And I guess we are trying our best to make certain pockets of the world better. But then at the same time, it's much bigger than just like a few hundred data centers.
[00:27:10] Matt: [00:27:10] Yeah. So when you go into a new country or expand an existing country, and you talk about your pops, you talk about your data centers.
[00:27:17] Do you use co-location space that other people have built, or do you actually build your own and deploy your own data centers? Or is it,
[00:27:24] Dalerie: [00:27:24] Oh, we typically use other people's co-location space that they've built. Right. So we rent the cabinets and the rocks, and then we put in our own equipment.
[00:27:31] Matt: [00:27:31] And how David, how do you imagine doing that?
[00:27:33] Because I imagine that some of the places that you may want to go to get to the 10 milliseconds don't even have that infrastructure, have you thought, is it gonna take a village or, you
[00:27:43] David: [00:27:43] know, like our current approach, it's a lot nimble than the big players. Like AWS. Like if I add up to us, they, we need, uh, like a lot of scale in order to decide to build a data center.
[00:27:55] Right? I
[00:27:55] Matt: [00:27:55] mean, they don't even look at anything that's less than a megawatt often.
[00:28:00] [00:28:00] David: [00:28:00] Well, by doing the call by renting someone else's space, having said all of that, to your point, if there's like a new market that we want to be the first, then that's a big, uh, it's a different problem for us to solve. Maybe the concept of.
[00:28:14] Like a micro data center may work.
[00:28:17] Matt: [00:28:17] Yeah, that's really, I mean, that's, that's how we solve it. Problem at vapor IO is we, we need to go to places where there aren't existing data centers. Now it's within an existing metropolitan area that has data centers, but not in the right places for these low latency workloads.
[00:28:28] And so we've had to deploy, uh, micro, modular data centers and we. Our most popular deployment unit. Now is these, uh, about the size of an SUV. It's a 20 kilowatt micro data center. And it's really interesting how fast we can deploy those. And so it's an a model I think that would potentially work for you in an emerging markets.
[00:28:46] Super interesting. I was
[00:28:47] Dalerie: [00:28:47] going to say, we could partner with you guys on that. Maybe
[00:28:51] Matt: [00:28:51] it wasn't suggesting that, but yeah, there's that, but it's very, it's interesting to see the similar and some of the problems, you know, it's like the problems that you're encountering [00:29:00] countering at the emerging markets writ large are similar analogous to the problems that we're encountering in a Metro market in the same way that you're extending the internet backbone.
[00:29:10] All the way to Jakarta. We're extending the internet backbone from downtown Chicago to North Chicago. And it's a, it's a, it's interesting that the, you know, the kind of fractal nature of that, it really fascinates me. So how do you see the future of cloud computing intersecting with this, these edge edge deployments?
[00:29:27] Dalerie: [00:29:27] So I think David would be really a great person to speak on that. Gartner has, you know, put together a lot of research around cloud and edge and how they're complimentary. And so cloud is always going to be here as the core, but as more and more applications are being consumed by users all over the world at the edge, you need the edge computing component to compliment that.
[00:29:48] David: [00:29:48] Yeah, I would definitely agree. I actually have a pretty interesting personal experience to share really at home. It's nothing kind of high tech. So I've got a cat, uh, whose name is [00:30:00] Oreo. Um, so he likes to scratch my sofa. So, and then, um, I've been kind of really pissed off, so trying all the way, like I tried all kinds of ways to get him off my sofa.
[00:30:11] But someone in the technology field, eventually I found like one solution. I too have tried to spray and then all of that first, but then the technology solution I was able to create was to have a rap camp facing the camera. And then if there's any kind of movement in front of the camera, then
[00:30:29] Matt: [00:30:29] shoots a laser at the cat
[00:30:31] David: [00:30:31] like that, like a little speaker playing a scary sound.
[00:30:36] So he will run away and that works. You've set this up the problem partially, but at the same time, it creates other problems. Now this poor little cat Oreo doesn't want to get anywhere near the sofa. I'm like fair to him, right? It's not like sometimes he's just walking around, uh, fairly innocent, but now he doesn't want to [00:31:00] get close the sofa.
[00:31:01] So it is a solution, but it's not graceful. Then I start to think about like what be a better solution to have. What if the camera has some sort of AI capability that can detect that the cat is scratching, then we will scare it off if not just pull cat along. But then now we are talking about like a real world problem with some agile AI capability.
[00:31:28] I think in the future, we should expect to see a lot of those real-world problems being solved by the combination of ad cloud. And then AI. So that will, can, people can really benefit from the Alyssa solutions.
[00:31:42] Matt: [00:31:42] That's a great example. I didn't, I didn't, I didn't know you were going to go there. That is a really good example.
[00:31:46] And it's such a simple example, but like, I would want that same thing too, if someone would make it cheap enough. And certainly, you know, you look at like how an Alexa works, right? Like it's like, you know, it's all the intelligence happening up in the cloud. Not as fast as I'd like it to, [00:32:00] so we need more edge, edge servers, but I can see you're right.
[00:32:03] Lots of. Lots of examples of things like that, where you just have these little nice, tiny improvements of intelligence that can be deployed ubiquitously because they're so inexpensive and they're leveraging shared infrastructure. And so on. That's a, that's a great example, David. So thank you, dowery and David for joining us here on over the edge.
[00:32:20] It's been a terrific time where, uh, can they, can people find you online?
[00:32:25] Dalerie: [00:32:25] Um, so at www.zenlayer.com and that's Zen, the N and layer as in layer of clothing, L a Y. And
[00:32:34] Matt: [00:32:34] how about the two of you personally? Are you on Twitter and LinkedIn? What are your, what are your Twitter handles? If you have those?
[00:32:39] Dalerie: [00:32:39] I'm not on Twitter, but I am on LinkedIn.
[00:32:41] So just my name or Dalerie Wu, I think slash Dalerie on LinkedIn.
[00:32:46] Matt: [00:32:46] Great. And David, how can people get ahold of you if they went to,
[00:32:50] David: [00:32:50] uh, just start for a David Xie?
[00:32:52] Matt: [00:32:52] Awesome. And that's X I E . okay. Well, awesome. It was pleasure to have, have the two of you here and, uh, I'm [00:33:00] excited to put this episode out.