vXchnge is officially tossing its hat into the podcasting ring with our new "Talking Data at the Edge" podcast. In this premier episode—titled "The Data Center: Getting the Distributed Infrastructure Right"—Technical Journalist, Ellis Booker, and vXchnge's Senior Vice President Sales, John Panzica, discuss the latest research into global IP traffic and some interesting use cases of distributed infrastructures.
Ellis: Welcome to the vXchnge podcast. I'm your host, Ellis Booker. Today's topic is The Data Center: Getting the Distributed Infrastructure Right. We're going to unpack some of the latest research into global IP traffic and trends and how this rapidly changes the data center requirements of companies. At the end of the show, we'll unpack some interesting use cases of distributed infrastructures. Joining me is John Panzica, Senior Vice President Sales at vXchnge. Hi, John.
John: Hello, Ellis. Looking forward to being here today.
Ellis: John, you and I were talking about Cisco's latest VNI global IP traffic forecast, released in May. The forecast has some mind-blowing predictions for IP traffic between now and 2019. For instance, they say that there will be nearly 3.9 billion global Internet users, more than half the world's population, up from 2.8 billion in 2014. And then also by 2019, there will be 24 billion network devices and connections globally, up from around 14 billion today. John, what else jumped out at you from Cisco's report?
John: Well, the numbers are certainly staggering, Ellis. If you look back at the history of the report, they started tracking IP growth about 1990. And it's been interesting because each prediction they make, it seems that technology and the use of IP has really blown out the forecast, and the numbers have always been higher. And it's based on a lot of factors of innovation and technology advancements across industry, residential, and it's staggering and it's really, really starting to accelerate as we move forward now.
Some examples of trends that you see behind it, for example, now, look at the residential market, or your home, and all of the smart devices that are Internet-ready, from your temperature of your home to your smoke alarms to your appliances and cameras. All of that information is accessible on a smartphone, and it's pushing intelligent data out to not only yourself but the service providers. So the sheer amount of data that now comes out of our home, or what they call a Smart Home, certainly contributes to that.
Simple other things in life, like athletes that will either cycle or will run, historically those are times to get away from technology and unplug. But through wearable technologies and GPS, you're seeing runners now with watches that give them all kinds of data on how fast they're running, what direction they're running in. And then the same on a bike, from a Garmin computer that will tell you where you are. And then, associated applications that will work in conjunction with it, like a company called Strava, where you can actually measure your time and riding your bike on a road versus a database of other people who have done the same route. All of that information is just data that comes off of either your bike or your wearable or out of your home, and it's factors like this that have really, really just incremented the growth of IP traffic.
Ellis: Along with overall IP traffic, the Cisco report looks at mobile IP traffic, right?
John: Oh, absolutely. There's some staggering numbers on that front as well. If you look at world population, there's 5.2 billion global users. That's up from about 4.3 in 2014, so a 21% rise. Even more staggering is looking at just the number of mobile-ready devices that are available today. It was only 4 billion last year in 2014. It's more than double, and it's 11.5 billion today. You look at things like this, it's not only your mobile device switches and smartphone, it's a tablet, which is also a television. The practical uses are really driving the adoption of these smartphones and tablets.
Uber is a classic example of smartphone adoption. And you think about that company, it's a multi-billion dollar company that's based on a service using smartphones. Well, individuals who want to ride, as well as the drivers, will possess an application, and they're actually able to transact and accept fares, pay for their fares, and also know where each other are, from GPS and where your car is to where the individual is standing who needs the car service. If you stand back and think about that, that is based on not only the mobile but also the amount of data, and we talked about IP traffic before that's transmitted. All of those things are factors in the adoption of smartphones, and there are many, many more examples of these kinds of useful and practical applications that affect our lives.
Ellis: Okay, John, what other trends are influencing application architectures and, as a consequence, data centers?
John: That's a very good question, Ellis. Being in the data center space of vXchnge, we come across a lot of infrastructure providers, service providers, that are in our data centers. And one of the biggest things is cloud and cloud computing and the different forms of it. What's interesting is there are a lot of folks who'll say cloud is taking revenue and/or services away from the data center. Well, in actuality, that is true, where a lot of companies are using cloud computing or infrastructure on demand, instead of going to the physical co-location site today. But also at the same time, cloud computing is the very fuel that is igniting the data center space.
Some interesting stats that were in that Cisco report, by 2018, more than 78% of workloads will be processed in cloud data centers. Twenty-two percent will be processed in traditional data centers. So those cloud workloads actually reside in a data center somewhere. And when you start to peel back the onion of the different components of cloud and the growth that we're seeing, some other interesting stats. So by 2018, 59% of total cloud workloads will be Software as a Service, which is up from 41% since 2013. So no longer are you doing on-premise software. It's a hosted software, which is a form of cloud, which is also in the data center.
Another stat on Infrastructure as a Service, 28% of total cloud workloads for Infrastructure as a Service will be in the cloud data center as well. That's up 44%. And the last one is on Platform as a Service, so the workloads for Platform as a Service today are around 15% and they'll be scaling to north of 20% by 2018. So some significant drivers in the cloud space and, as well, fueling the data center market.
Ellis: So we have this phenomenal growth in IP traffic plus mobility, plus virtualization, which all fuel in growth in distributed infrastructure and the need for data centers, as you pointed out. How does the provider, like vXchnge, reacting, John?
John: That question is actually at the core of our value proposition when you're talking about distributed infrastructure. And if you look at some of the trends that I was talking about prior, with Garmin and Strava or Uber, it all revolves around users that are in different geographic locations, as well as the companies themselves. Uber's headquartered on the West coast in San Francisco. However, they have users all over the world. And if you think about practical application-wise, that if you have users that are sitting here in New York, as an example, and they want to interact with Uber, their experience is based on their response and the friendliness of the application that they have on their smartphone.
So there's a lot of data that gets transferred between the Uber systems and the smartphone and the driver's smartphone. And the really use of that, the data has to come in quickly. There can't be a lot of delays. It has to be user friendly. The GPS has to work timely. All of that actually boils down to a common denominator of where infrastructure is to support the application performance, as well as the user's experience with the application.
Ellis: So as I understand, John, it's a latency of the connection, which you can't get around. It's basic physics, right? And the user experience that requires these companies, like Uber, to have facilities nearby where the applications are actually being used. The person in the cab doesn't want to wait to get the server in Hong Kong to respond if there's too much of a latency. Is that what's going on?
John: Yeah, that's exactly what's going on, Ellis. So the ability for the user to get timely responsiveness from the systems that are connecting the driver and the individual together and all of the critical information, getting information to them quickly. So it's latency and making sure that the infrastructure is closest to the user as possible. And that really goes to the core of what vXchnge's value prop is. We're actually deployed in 15 data centers across the United States today.
You may have read that one of our most recent data centers that we've opened up is in Philadelphia. Some folks have been like, "Why Philadelphia as market?" And if you sit back and you look at some of the statistics behind it, Philadelphia is one of the top five cities in the United States. And you look at the geographic proximity of historically how Philadelphia was served, a lot of the infrastructures being in New York and a lot of the infrastructures being northern Virginia/D.C. area. And because of mobility, because of applications, like Uber and Strava, applications that I've mentioned, you have to worry about latency and performance and user experience, which means you need to decrease your latency and you need to have infrastructure that sits local in those markets.
So no longer is it acceptable to serve Philly from New York or from D.C. You have to serve Philly from Philly. And a lot of service providers are moving into data centers because of those reasons in those types of markets.
Ellis: John, I promised up top that we'd talk about some real world examples. Can you tell us about some companies who are leveraging these trends in interesting ways?
John: Sure. Absolutely, Ellis. So I pulled a press release out from a company, SAP. It's a German software company and they have a host of software products that work in the cloud, as well as big data analytics programs called HANA that do in-memory processing. And they just recently announced an agreement with the National Hockey League, the NHL, and they were going to provide not only technology services, but the NHL was going to actually launch some new applications. And it was around taking the data that surrounds the NHL.
And if you peel back the NHL for a moment, the league was formed in 1917. I did a little bit of research to understand the magnitude of this. And there are 30 teams with 23 active players, and then there's usually about 50 that are on contract with the team. And if you think about that, going back to 1917, the amount of players there are, then the amount of teams that there are, and then look at some of the stats that they're trying to capture and record, again. So how many penalties there are in a game? Or how many off sides there were, how many goals scored, how many minutes per player on ice? Variations of that across it, there's probably 30 to 50 different metrics that are captured.
So if you put all of that together and you multiply it times the history of the league and the team, we're looking at tens of millions of records that change on a daily basis, on a minute-by-minute, a second-by-second basis actually, as you watch the games. And to be able to have NHL fans sit there with an app and pull up online stats that are real time updated on their players or a game that they're watching on T.V., it's staggering. It's staggering the amount of data and that works around big data analytics.
And SAP is a great technology provider. One of their applications, HANA, which I spoke about before, it's essentially a big data analytics engine. And they're processing all of this data for the NHL in-memory, so they're not storing it to disk. The HANA platform processes in-memory, so it actually decreases IOPS, as well as helps with latency and getting the users the data in real time.
So you think of the magnitude of data that these fans are pulling and where they're located throughout the country, throughout North America, throughout the world, for that matter. The geography and the amount of data is just staggering. And that's a classic use case of needing distributed infrastructure for a sports fan. And there are many other examples of that, but that was one that I felt was really fun and I wanted to share.
Ellis: Thanks, John. That's all the time we have for this episode of the vXchnge podcast. If you'd like to hear other podcasts or suggest show topics, please visit vXchnge at the link that appears above. And please, subscribe to our podcast.