Il contenuto dei post sulle risorse viene visualizzato solo in inglese.
Man: Welcome to “Not Your Father’s Data Center Podcast” brought to you by Compass Datacenters. We build for what’s next. Now here’s your host Raymond Hawkins.
Raymond: Now welcome today to another edition of “Not Your Father’s Data Center.” We are joined today by Phillip Marangella, chief marketing officer at EdgeConneX. Phillip’s got a great networking background, great data center background, and gonna provide us some insights on what the edge means, what the network means, and how they’re coming together and how it influences the rest of the data center industry. Phillip, thanks for joining us today.
Philip: Sure, Raymond. Thanks for having me on today. You know, I’m the CMO at EdgeConneX. I’ve been here for a little over four years. And I’ve been in the data center space for over 10 years, prior to that, been in the telecom space. So really kind of span the spectrum in terms of network and IT and data centers and so forth. And, you know, since I’ve been at EdgeConneX really it’s an interesting time, you know, over the last few years in terms of the advent of cloud and now we’re on the cusp of all these other trends. And then, of course, today, you know, we’re confronted with the pandemic so that that creates an interesting scenario for us as data center operators. So again, I’m pleased to be here today and look forward to kind of talking a little bit more about some of these topics and issues, Raymond.
Raymond: So thank you for mentioning the pandemic, Phillip. Just so folks for point of reference, Phillip and I are recording today, I’m not sure when you’ll get to listen, but we’re recording day Thursday, April 16th. And our country, you know, is in the throws as everyone will know of COVID-19 and what it means and all of us experiencing the challenges around our children at home and how much bandwidth is getting used and how fast is Netflix download, and all of those other questions. Do our Zoom conference calls work properly? I think that the world’s getting a little bit of a glimpse as the bandwidth goes up, what could happen for lack of a non-technical term the backbone of our network. So I think our conversation today fit nicely into where the market might be headed and then our current COVID-19 situation might be a glimpse into that. Is that a fair understanding of, hey, we’re just getting a little taste of what it could look like when the network gets gummed up?
Philip: Absolutely. Even before this occurred, right? You know, I mean, you talk about, look, I’m at home, I’m recording from my basement, you know, my son is probably playing “Fortnite” in the next room and my wife’s watching Netflix and my daughter is doing online course for her university, so we’re all, you know, heavily pooling from the network. And but even before this kind of new situation, you have so much content, so much data that’s being consumed at what, you know, I would argue is the edge, right? Be it at home, be it in a lot of these secondary, tertiary markets, and so forth. And a lot of content’s being created there, not just consumed. So, you know, the internet was always, I like to talk about, you know, us in general as data center operators helping really rearchitect the internet and because of all this data gravity that’s occurring and shifting to the edge is creating a change in these network traffic flows because traditionally it’s been download centric, unidirectional and now it’s becoming, you know, at the edge content, staying at the edge, going back to the core. It’s becoming this multi-directional nature that requires these edge data centers to interoperate with core data centers and to help alleviate these traffic flows with things like autonomous vehicles and machine learning, right, internet of things, and on and on. They’re all in their own right creating a huge burden on the network and this is the role I think a lot of the data centers will help alleviate and have alleviated, particularly in the current situation with COVID as the sudden burst of traffic flows at the edge.
Raymond: So, Phillip if I can, so I like to try to make things a little simpler. So with your background, especially your background, going back to Nortel and your time at Verizon, I think you have a pretty rich network understanding. Can I repackage something you just said and just make sure that I get it right? I think what you said is we’ve largely been a download network environment. And what I think that means is I’m a user, I’m sitting at the edge and I need to download a movie, I need to download an image, I need to download an application, I need to download a file and that’s my primary interaction with a network is I needed to download something off the network to my machine to work on it locally or to view it. And today it’s much more of I’m sending requests, I’m getting services back on the other end of the network and that those services are bringing some answers, some query, some question back to my machine and the traffic now going both directions is far more important. Is that a fair layman’s way of saying what you were summarizing?
Philip: I think so, Raymond. I mean, the other point is the download nature, right? The volume of that download and the velocity, right? The latency sensitivity of these downloads are dramatically increasing, right? On the one hand. And then on the other hand, as I said before, there’s so much data that’s being created at the edge. You know, think of all these TikToks, now my son and family is bored and they’re creating TikToks. That’s a lot of content being created at the edge that’s going back to the core, right? That’s just one minor example, right? And, you know, you look at smart cities and you need this kind of much more distributed architecture of network and data centers and applications, how they interoperate amongst each other to alleviate these bottlenecks is what we’re really trying to solve for and what you were talking about in the beginning.
Raymond: This idea, I like you made the comment, Phillip, you said that we’re really rearchitecting the network or the internet. And I think the theme of today’s show was to talk about the edge and COVID-19 jumped in the middle of our conversation and hey, we’re getting this glimpse into performance. But really that’s what we’re talking about is what does edge mean? And, for me, you know, you have a great network background and technology as well. I came from the technology space and we always talked about you needed to distribute compute for a while and then eventually you consolidate compute and we go through these cycles of what’s most important? You know, do we need to consolidate for security and control purposes? Do we distribute for access and performance purposes? And that those competing needs are always pulling or pushing against each other. And today I think with the edge, we’re clearly in that distribution mode, right, where the network demands are saying, hey, we need to distribute some of that compute. And I like your TikTok example, right? Not a business-related app but certainly a user experience of I’ve got this, I don’t know how long those videos are, 8 or 10 seconds, maybe 30 seconds. And that content is getting delivered but it’s getting then pushed into the network and where does that go and how does that performance need to work just as an example that I think everyone at least with a teenage to up kids can understand going on.
So edge, Phillip, I’m gonna be a bit of a long preamble to this question. So I remember when cloud was just something we talked about and I would go meet with customers and they would say cloud meant different things to each customer and they were all very concerned about it and working on it but didn’t have meaningful projects and that’s changed over the last 15 years. I feel a little bit like we’re there with the edge again, that all of our customers are talking about it, studying it, thinking about it, have teams looking at it but are still working on what’s the right business case, what’s the right definition, what of their customer’s needs need to be satisfied with the edge. You dropped a couple of keywords, you know, you talked about autonomous driving, machine learning, and Internet of Things. With edge being in your business’s name and EdgeConneX, can I get you to tell us how you guys think about the edge and with the caveat that I know customers’ environments are different and everyone’s situation’s different, but when you say edge, as the chief marketing officer at EdgeConneX, how do you think about it? Help us understand that and explain that a little to us.
Philip: Sure, Raymond. I mean, look, and I’ll say like it’d be good to hear from you as well, Raymond. You know, we both have common customers, and, you know, kind of myriad of requirements, right? And I often get asked that, right? You know, we’ve been at the edge game for many years now, but at the end of the day, it’s not how I define it, ultimately it’s how the customer defines it, right? I’m not gonna say, hey, it’s two megawatts in a market like Las Vegas is the edge, right? Yes, we built a data center there for customer requirements related to content distribution but, you know, for that scenario, that was the edge, right? In another scenario, we’re in Amsterdam where we have a campus approaching 100 megawatts and for that cloud hyperscaler that was an edge requirement for them, right? That’s, you know, because of the existing capacity or capabilities wasn’t optimally located where the proximity to their customers or their ecosystem wasn’t satisfactory so they wanted something of that magnitude. And it varies, right? It could be for autonomous vehicles, micro-edge where you’re taking into considerations of ultra-low latency, sub-five milliseconds, right? Or that you have to be extremely close to where the compute needs to occur and where the data is being pulled from and that could be sub-100 kilowatts, right? It could be in a closet, it could be in a tower and it flexes, right? But the key is when we talk about those things like IoT, autonomous vehicles machine learning, cloud, they’re not independent, they’re all interrelated, right? All of them are leveraging the cloud, right? The sensors are going back to track it and trend it over time. And so this is where the network also comes into play of connecting the data centers, right? There was a Gartner report that said the edge will eat the cloud. And look, it’s provocative, I get it, but ultimately it’s not mutually exclusive, the two you have to interoperate. And so that’s why we talk about hyperlocal to hyperscale and being able to fulfill those customer requirements for their edge, wherever it may be, whatever size it needs to be, and so forth.
So I go back to you Raymond, I mean, I don’t know if you see the same thing. I mean, you have the same in a way the model, right, in fulfilling a lot of tier two markets, hyperscale as we think about the cloud. I think yes, it’s gonna get closer to the edge and proximity matters, but it doesn’t mean the server farms go away, it doesn’t mean Ashburn goes away, the two have to work in conjunction.
Raymond: [inaudible 00:11:26.626], Phillip, I like when you said hyperlocal to hyperscale, that’s a great comment and I think you’re 100% right. The provocative idea that the cloud…or that the edge is gonna eat the cloud I think is just a fundamental misunderstanding of what’s going on. I love your explanation and I think you’re 100% right and I’m gonna try to tie some real-world applications to what I hear you describing. So you talk about IoT, an interesting example. So a friend of mine was buying a new refrigerator and the manufacturer of the refrigerator offered him a $250 rebate on the refrigerator if he signed an agreement that allowed their IoT sensor in the refrigerator to download data back to them whenever they would like. And so my friend calls and he’s like, why would they give me $250 worth of credit towards this refrigerator? I don’t care if my refrigerator connects to the internet. And for me that IoT connection for that refrigerator is all about them doing predictive analytics on the performance of the refrigerator. If they can get sufficient data across a portfolio of refrigerators and decide that they can shrink the compressor on that refrigerator by 20% without impacting the ability for it to cool, the amount of money they could save by that I think is just one example of IoT. It’s an internet of things, it’s a sensor at a refrigerator but that data in and of itself at that refrigerator means nothing. That data across 30,000 refrigerators back in the hands of a refrigerator manufacturer making a decision about a compressor, now it’s an incredibly valuable piece, but it had to make that journey back from the refrigerator to be meaningful. And that for me, I think that’s a business example of exactly what you just described, Phillip, about how one’s not gonna eliminate the other there, the value and the purpose and the meaning of them is closely related. I think that’s an interesting example. I just want to see if that resonates with you, Phillip.
Philip: Absolutely. And I mean that’s like 30,000 refrigerators but you’re talking about billions of different devices all going back to try to track and trend it to do predictive preventative analytics. Another example that I’ve had in the past, and speaking to GE and aviation, some of the, you know, senior executives there, they’re like, hey, one jet engine, you know, generates over, you know, half a gig of data on a single flight and at any one time there’s 40,000 to 50,000 flights occurring, right? So it’s a ton of data that’s being offloaded every time the plane lands. And you have to quickly analyze that to say, hey, before it takes off again, do we have to do any maintenance, right? Or do…you know, if a flight is typically going to the Middle East where there’s a lot of dust or if it’s going to China where there’s a lot of pollution or what have you, you might change your thresholds and you’re able to analyze that stuff. You know, there’s that where autonomous vehicles like vehicle to vehicle kind of communication, right, that’s where the latency comes in to be able to, you know, share what you’re…that car ahead of you seeing so that you can prevent yourself from having an accident or you know, avoiding a pothole or whatever it may be.
So there’s like all these use cases that are coming that are just creating this massive tsunami of data and how do you route that traffic? You don’t want to send it all the way from the car that’s on, you know, in Texas, right, to Ashburn and then back, you want to keep that local, you want to keep it approximate, right? But you do the trending data, will go back to the data lakes and so you can analyze it over time. So that’s the key role is to help efficiently deliver the traffic, right? And that’s where the edge peering comes into place in, you know, determining what stays local and what goes back to the core.
Raymond: Great stuff, Philip. And I appreciate these examples. I think that’s one we hear a lot, right? When people talk about the edge, you hear them talk about IoT, you hear them talk about autonomous driving, you mentioned machine learning, I think there’s lots of other things that we think are coming. With your network background and telco background, could you take two minutes… You know, we hear the word 5G a lot, you know, I think a guy like me who doesn’t have a network background that you do, I hear 5G and I just think my phone’s gonna get faster. As you think about 5G and what it means to what’s gonna happen closer to users, can you give us two minutes on just your thoughts and how it impacts us doing what you said early, rearchitecting the network?
Philip: Well, you know, so 5G often comes up, right? And, you know, when is it gonna happen and so forth. And so as a marketing guy, right, a lot of the carriers have done a good marketing job and saying, hey, 5G is here and this is what it’s gonna do. You know, the delta is there, right, between the marketecture that they put out and the reality of what it can and can’t do. Yes, when it comes, it will allow greater network speeds and capacity and all this kind of thing. But that in itself also creates a burden, right, in terms of, hey, great, great now that you can do more, you will do more, right? And that just creates further content and the interplay between the wireless and the wireline networks will come into play in terms of how do you backhaul traffic and so forth. I think, you know, it will be great when 5G comes, but don’t forget that the wireless networks have invested billions upon billions in 4G and so they’re gonna look to optimize that as long as they can. A lot of the workloads and applications that are hitting those wireless networks, 4G is still plenty good enough. And so, you know, it’s kind of metering out the 5G deployments and growth to align with those applications and workloads as they come to bear.
Raymond: Phillip, you raised an interesting point. I was at an infrastructure investor conference recently just to give people an idea of how interesting my social life is. And they were talking about this 5G and they were an infrastructure investor, so guys that invest in bridges and highways and, you know, dams. And one of the comments they made, which I think just rings so true with the comment you just made about the carrier’s investment in 4G, they’re trying to continue to maximize the return on that investment. And I’m not gonna get the number right, but the idea is it was really big. This investor infrastructure fund said, “Hey, we’ve done an analysis and for us to roll out 5G globally is $7 trillion.” And he just said, “Hey, I just want everybody to stop.” He goes, “I’m not a technician, I’m not a network architect, I’m not thinking about this from a throughput and capacity standpoint, I’m just saying, guys, $7 trillion is a lot of money and where does the capital come to do that? It doesn’t come in a year or two.” So when people talk about my iPhone is gonna go from 4G to 5G, that’s really the beginning of the process, isn’t it, Philip? I mean there’s all of that backend infrastructure for us to fully enjoy the benefit of 5G, that $7 trillion number, maybe it’s high, maybe it’s low, I don’t know enough to know. But the point being is it’s many, many, many years of investors laying out capital to build that infrastructure. And I think that ties into your 4G comment. Is that a fair assessment that “Hey, we’ve got a long way to go?
Philip: Absolutely, right? So I used to work at Verizon business and not to get in specifics, but just say like, “Hey, what if it costs $7 billion for them to do their own network, right? And just, you know, 1% of what you’re talking about, right? You know, their capital budget every year is a few billion dollars, but it’s spread across hundreds of products and so forth. So the time for it to take a full robust roll-out, it’s certainly gonna take time and so they’re gonna be selective. It will take time and in the meanwhile, you know, they’ll try to optimize 4G. But again, this is the interplay, just like from the core to the edge, it’s also the wireless to the wireline. And this is why this is cool, this podcast that you’re doing Raymond, and I appreciate it, right? Because you think, hey, why are two competitors coming together, right? Because look, for us, there is so much demand from a data center perspective, there’s gonna be so much demand from a network perspective, but we can’t work in silos, we have to collaborate and work together to solve for this to rearchitect the internet as we’ve been talking about, right? And how do the wireline and the wireless carriers collaborate with carrier-neutral data center providers to help create these pairing infrastructures at the core, at the edge, and to really help the chocolate flows for all these new applications of workloads that are coming?
Raymond: So, Phillip, you brought it up that, you know, we’re both in the edge business and we’re both in the data center business, we both build largely single-tenant buildings and that I think many would view us as competitors. And so this is a bad analogy, but I’m gonna go down the path anyway and I think it’s a way to think about our business, is I was watching the news this week and Sonny Perdue the head of, Secretary of Agriculture was trying to help out dairy farmers because the dairy farmer demand has gone down 25% due to COVID-19. And when I think about that, that industry is down 25% but what it represents is a bunch of different dairy farms, right? And I don’t see those farms going, wait a minute, I want to make sure my farm sells X amount more gallons versus that farm, it’s the industry is hurting. And that’s a poor analogy but where I’m going with that is mine and your industry is the exact opposite. I think all of us are so busy, we’re all serving our customers and helping them, and we’re all going as fast as we can that we have the opposite problem of our friends in the dairy industry who they’re down 25%. I think what’s going on right now is our industry, especially with this experience around COVID-19 is recognizing how vital the network that we all help support is. And there’s so much demand for mine and your business that there’s enough work to go around and there’s enough challenges to go around and there’s enough tough problems that us, I think you’ve used the phrase I’ve heard you use before, that we’re, we’re really in “coopetition”, right? Occasionally we compete with each other, but the reality is we’re helping industry grow and we’re helping a network grow and we’re helping solve a problem. You know, I know we’re not curing cancer, but we’re helping solve a problem for society, right?
Philip: Yes. I mean, we’re really enabling the digital economy, right? We’re key… Like what we do isn’t sexy but what our customers do is, right? And some of our customers support healthcare providers that are coming up with cool R&D that might help cure cancer, right? So that’s what’s really cool, right? But you gotta we have to work together to bring together the infrastructure and that’s why we’re so important because we are critical infrastructure. I’m very grateful to be the business that I’m in and it’s heartbreaking to see what’s going on in the news, not only with the people suffering with COVID but then the people who are impacted economically. But, you know, I think though, you know, that’s why being critical infrastructure, it’s really important that we work together, we collaborate to kind of solve for these problems so that, you know, people can work from home, people can create new jobs, right, in the digital economy that will offset some of these service jobs that are lost. So it’s challenging times but, you know, certainly from a technology perspective, from a data center perspective for the foregoing, you know, 5, 10, 15 years, there’s an unlimited demand I see coming down the pipe.
Raymond: So, Philip, I’m gonna date myself a little, and hopefully this resonates with you. There was a…BASF had a campaign that said, we don’t make the products you buy, we just make the products you buy better. And as my team here at Compass talked about what it is we do and you alluded to it a little bit, I’m like, guys, we’re not curing cancer, we’re not securing the financial networks of the world, but we’re the infrastructure that sets all that up and I don’t want to steal from our friends at BASF, but I think that’s a lot of what I hear you saying is we’re enabling, you know, the transfer of information and the exchange of ideas and the infrastructure, what we say [inaudible 00:24:24.146] we’re trying to provide a secure place for every industry to plug in, that’s it. Just let’s give them a safe, secure place to plug in so that they can do the things they need to do, whether it’s come up with a vaccine, whether it’s help manage people’s finances, whether it’s helping kids get into college, whatever it is, we want to provide a secure place. And I think there’s enough of that demand that you and I both get to work for exciting fun businesses and an industry that’s growing like crazy even in these tough times. And I like you feel super fortunate to be in that space.
Philip: I know exactly that commercial you’re talking about and I reference it often because they hit the nail on the head. So, you know, we’re the quiet guys behind the scenes but what goes on in the servers that reside in our data centers is some really cool stuff.
Raymond: Well Phillip, if you’ll let me, I know we only got a few more minutes, I’m gonna go down one more branch and see if we can talk around this for a minute. We talked a little bit before or at least I think mentioned the idea of tier two and tier three cities, especially in the concept of the edge. So you and I are both on the same page, that edge means something different to our customers, I loved your 100-meg example down to your 100-kilowatt example, I think you’re completely right in that analysis. But let’s talk about big data center towns, big metro areas, I think everyone gets those right? I think an edge data center in metro Atlanta or metro New York, metro Dallas or LA or Chicago, service provider trying to provide services in that metro may need several nodes, but I think that tier two and tier three city as we talk about building the infrastructure to build out the network, I think that tier two, tier three, and I’ll go so far as to say that tier four-city, where I’m going with this Phillip is I think back, I know this is a silly example, I think back almost 100 years ago to the rural electrification commission when the government in the ’30s and ’40s said, “Hey, we’ve got to get electricity out to western Nebraska.” And it’s not economically viable for a business to do that but they need electricity. I think we are gonna see the same thing in mine and your business, that high-speed internet services are gonna be needed to be out there, but there may not be an economic incentive for a business to put, you know, in eastern Washington or western Nebraska, that kind of a thing. And to me, I’d almost qualify that as a tier four-city. Could you just give me two or three minutes on your thoughts around the tier two outside of the big markets, tier two, tier three and maybe even tier four?
Philip: Sure. I mean, you know, we’re in a lot of those markets, I think a great example is Portland, although that’s in its own right becoming quite a big market, you know, for various reasons but, you know, we’ve been there for a number of years. And, you know, it’s interesting, you know, Amazon came to us maybe three, four years ago and put its direct connect node in there. Because keep in mind, you know, if you wanted to get to the cloud, traditionally you’re gonna go up to Seattle or down to Santa Clara, but putting a direct connect node there, giving customers that private dedicated access, lower latency capability really helped drive growth and demand for cloud compute services locally there in Portland, right?
And, you know, another example is from a content perspective. Phoenix was a great example for us where, you know, a large MSO cable operator would traditionally service that market out of LA, but the cost of serving it, of transport and the performance of the content that they were serving was just so much better to do it locally there and serve it out of a Phoenix data center, right, and put their cable head ins there. And so you see on and on and now we are going to South America, right, traditionally that was served out of Miami, maybe Dallas or perhaps LA, there was a large concentration in Brazil to kind of support all of South America, now we’re seeing more and more customers want to kind of distribute across South America. We’ve built in Buenos Aires, we’ve gone down to Santiago, will likely go to a few markets, right, there. And Europe going beyond the flap, right? We’re in Munich, we’re in Warsaw, we’ll continue to expand there, right? So you just see this kind of, it’s not just in North America, right? But now you’re starting to see for these major service providers to have this globally distributed architecture to become more proximate to the eyeballs, to the enterprises, and to the consumers that they want to kind of serve. So, you know, absolutely, right? And the importance there too, Raymond, and I think you guys have a great story as well, is the speed to market, right? Because you got to have this kind of modularity so you can quickly go to a new market, bring up a data center quickly, bring in the ecosystem of the networks, and so forth. But, you know, for us it’s important to collaborate with our customers to give them exactly what they want, where they want, when they want? And the speed is an important factor.
Raymond: No question, every customer meeting we have, Phillip, we get asked how fast can you get there, right? We talk all the time, customers really ask us three questions as qualifiers and until we say yes to all three it’s are you in the market, you know, specifically, Northern Virginia or Chicago, are you in the right market? Do you have the capacity I need when I need it and can you handle my growth when I plan to grow? If you can’t say yes to those three, they’re onto the next provider. And so no question, how much and how fast is early, early on.
Well, Philip, thank you so much. I know I appreciate you telling us early on at home with the kids as we suffer through this pandemic and recording from your basement, we appreciate you doing that and appreciate just talking a little bit about what the network means and how our industry supports the growth of the network and how edge means something different to everybody. It’s been great having you, and I appreciate you taking the time to talk with us today, Phillip.
Philip: Awesome. Well, I appreciate it as well, Raymond. Thanks for having me on, enjoyed the conversation very much today.
Raymond: Sounds good. Take care, Philip.