October 4th, 1957. The Soviet Union launches Sputnik into orbit, and with it, a perilous escalation in the Cold War. The Defense Advanced Research Projects Agency is formed with a mandate to keep our military and technological assets ahead of the Soviets. From his tiny office in a corner of the Pentagon, one engineer not only recognizes the need to rely on more powerful computers to win this conflict, but that these computers must somehow find a way to “talk” to one another.
In the first episode of Traceroute, we look at how the need for interconnection shaped the fundamental building blocks of what would become the internet. From ARPANET to the Telecommunications Act of 1996, what started as a military exercise grew to reflect society’s basic need to feel and stay connected. And with that growth came an unprecedented need for hardware solutions that had never been thought of before. With insights from Jay Adelson, Sarah Weinberger, John Morris, and Peter Van Camp, we’ll look at the social, political, and technological forces that started the internet.
JAY ADELSON: [00:04:11]
A lot of people assume is just sort of magical, like it's air that we breathe or electricity that just happens to be available to us. But in order for everything to continue to grow and reach its ultimate capability. We need systems and technologies and businesses that allow that infrastructure to grow.
GRACE: This is Traceroute [rowt]...
the show about the inner workings of our digital world - all the physical STUFF that you never think about. In a world that feels intangible.... We look at the real people and services building, maintaining, and scaling the internet.
I’m your host, Grace Andrews. a technical storyteller and product evangelist at Equinix.
In this episode: INTERCONNECTION.
Act 1: The breaking point
Today’s internet exists not just because of plucky tech entrepreneurs in Silicon Valley... Or successful Wall Street IPOs... It exists because of infrastructure. The roads and bridges built over decades to allow the internet to function as it does today: We take a lot for granted –lightning fast download speeds on our smartphones, thousands of family pictures being stored in the cloud... and well, the ability to do a lot of our jobs remotely for the last year...
Obviously, this took a lot more than a few lines of code. In this series, Traceroute will look at the stack - the different parts of internet infrastructure. We’ll talk about how they were developed, why they matter ...and why continued open access and interconnection is vital to the world as we know it.
there was a fairly sophisticated group of engineers who said that the Internet could never scale and by the end of 1996, it was just going to collapse under its own weight because it was just too distributed.
That’s Jay Adelson. He’s a serial entrepreneur, currently focused on his gaming company... But back in 1998, he helped to start a company called Equinix - a neutral internet exchange carrier. The internet was on fire....but Adelson saw trouble on the horizon. Companies and tech founders were creating little corners of influence, They could settle into these corners and close them off to outsiders....they would dig in their heels and fight for users....it was like a million different islands with zero connection between them. But how did we even get to this point?
Act 2 - DARPA and the Early Need for Connection
DARPA was initially created as a reaction to the Soviet Union's launch of Sputnik in the fall of 1957. the Defense Advanced Research Projects Agency, born during the cold war, is ironically the moment when the U.S. starts learning the value of interconnectivity..
Sputnik was the world's first artificial satellite and it created eventually a panic in the United States, and a fundamental reorganizing of national security agencies// It was, in essence, the nation's first space agency before the creation of NASA//But it was also given this very broad mandate to take on those research projects as directed by the secretary of defense. So this idea that it would try to create new technologies even beyond space, to keep the Pentagon, the military ahead of the Soviets. Sharon Weinberger is a national security reporter and editor. She’s spent 20 years focusing on the intersection of science, tech, and national security. And she wrote a book called The Imagineers of War: The Untold Story of DARPA, the Pentagon Agency That Changed the World. Sharon says DARPA initially focused on what you might expect - space programs, rockets, missile defense.
But it was also thinking about communications...
It was looking at command and control of nuclear weapons because, of course, this was what was really the threat of Sputnik and the Soviet Union was the concern that the Soviets, if they could launch a satellite into space, they could launch an intercontinental ballistic missile that would reach the United States from the Soviet Union very quickly. So THEN, we get the birth of ARPANET - the precursor to the internet. A network where computers could communicate with each other. [WEINBERGER: 00:10:56 This is still sort of early days of computers. // DARPA was being sort of saddled with an old air defense computer big large white elephant. If a computer that was part of command and control and told to do something with it] So DARPA hired JCR Licklider An internet hero that you may have never heard of. Licklider was a psychologist. Before working at a federal agency, he was in academia...doing early computer experiments at MIT.
So when he came into DARPA what he said is let's step back for a moment and think more fundamentally about how humans inte ract, how they interact with computers, he was really forward thinking. He sort of looked ahead and said, the way that we work with computers is going to fundamentally change. our society] Eventually the proposal became a prototype... and a handful of nodes became the first ARPANET network. In 1969, a researcher at UCLA sent the very first message over ARPANET to the Stanford Research Institute.
the very first message was low. But I remember hearing about the first time like, oh, maybe it was going to be low and behold, like some sort of historic message. And actually what it was was log-in and it failed after the first two letters. By the 1970s, we’d have something closer to email messages. But the user base was still limited....a few ARPA headquarters, some universities....just a few nodes. But after years of innovation - the network did start scaling. And that worried some people.
And DARPA, as an agency, like all parts of the Pentagon, was under a lot of funding pressure because there was a lot of blowback from the Vietnam War. You know, there was a lot of pressure for the military to only fund things. So why are they getting into social science? And so the DARBA director at the time joked, you know, he was looking around for projects to kill like, oh, computer networking, you know, what does this have to do with waging warfare? And he almost he's like, I was the man who almost killed the Internet because, you know, he was about to pull the funding] And whom exactly was this technology FOR?
[00:32:52] one of the real challenges to the early ARPANET was convincing the nodes, these research institutes and universities that this was with their time worth their time because people so jealously guarded their access to computer time. It was a really valued asset
[00:34:46] And so one of the examples that Licklider gave this guy interviewed was, you know, you could be in your kitchen and you could don't you don't think the word he used was downlow, but you could access recipes from a library.
[00:35:00] And I remember the guy who was interviewed said, well, you know what? What I what I want to do that for? He just couldn't understand or at least that was the thing he could grasp on to. OK, that's something I could use it for. the magic of what Licklider did and his greatest contributions to the Internet, which I don't think are credit enough today, was this real messianic view he had of trying to get people to understand why you would want computers to speak to each other
Over time, it made less and less sense for this new technology to be kept under lock and key by the government - they didn’t really need it for defense systems, and there were a lot of smart people realizing that the interconnected systems could have broader applications. By the 90s, the internet would be completely phased out of DARPA, becoming commercially viable and consumer-friendly. But the military legacy can’t be overlooked.
It accelerated this work possibly by decades, because there was no other agency in the world like DARPA that could, you know, the initial conversation to fund ARPANET in the mid 1960s was something like, you know, a 15 minute conversation with a director said, sure, OK, you can have a million dollars at the time was a lot of money for research. So, yes, we would have had a computer network, but it would have been later. And of course there was the unsung hero behind it all....
We tend in American society, especially these days, to love, you know, the you know, the the you know, the Steve Jobs that Elon Musk. J-C-R Licklider is not a household name today, and that's because it's not called...you know, he didn't start the Licklider computer Internet company.
[00:44:20] But the ARPANET is an example of this interaction between government research and the private sector and the fact that it's rarely one person who is the sole inventor or key to something success]
Act 3 - The FCC Will Let Me Be
But how does the internet go from a Pentagon-centered closed tool to an open web? First, it spreads to more universities and becomes an important research tool. Coders start creating new languages and protocols. Before long, people have internet in their homes - tying up their phone lines. By the 90s, we’re becoming more and more connected. And it’s time for the regulators to step in.
I had gotten onto the Internet back in the 80s and and that was more just as a as a curiosity. I've been playing with computers since the 70s, John Morris is a lawyer who’s been involved in internet content issues for decades.
[00:12:44]. back in the 80s.commercial communications were prohibited on the Internet, the Internet was only for a government and academic communications,And so the critical evolution really started in the late 80s and into the nineties is that the government tried, decided to essentially devolve give up its control over this Internet The unique decentralized nature of the internet made it easy to scale.
[00:06:15] back in the 90s, you know, an Internet connection would, you know, maybe just get to a developing nation in Africa? You know, I'd say Kenya was one of the early, early countries on the Internet. But if a neighboring country wanted to connect to the Internet, all they really had to do was to figure out a way to connect to Kenya. They didn't have to do anything to connect to the rest of the world. If they could if they could find someone in Kenya who would allow them to connect through, then then a whole new nation could come. Come on and that was really the difference that set the internet apart] Interconnection in the most basic sense, says Morris, is when two networks talk to each other. But as more people got onto the internet, those connection points became congested.
[00:08:24] if they're only interconnected in a single place in the country, then any traffic, any communication going from one network to the other has to flow through that one single place. MORRIS:00:09:37] AOL was an early email service provider. And amazingly, all AOL email went through Reston, Virginia, even into the 2000s. And so if an AOL customer in Argentina was sending an email to an AOL customer in Ecuador, that email would go through Reston, Virginia, which means that essentially the AOL network was not interconnected in a in a granular level. It was really a fairly centralized email system. And that actually had enormous implications for security, government surveillance, all kinds of things. That email was all flowing through Reston, Virginia Networks grow into an increasingly complicated web. and And then the rules of the game changed and growth exploded.
[00:15:03] that's the day that one could set up business on the Internet. And so it's not really that, you know, kind of setting up business on the Internet really transformed the Internet as much as it opened up complete new business models that could support all of the services that we know today The commercial growth continued, and then came the Telecommunications Act of 1996...
[00:16:18] was a massive, sprawling piece of legislation. Its primary focus really was to try to generate more competition in the telecommunications market. I mean, more competition among phone companies. AT&T had been broken up. Yeah. And it was broken up into a bunch of little kind of tiny AT&T. So you had and Bell Atlantic and which became Verizon and you had NYNEX and you had Pachelbel Bell and you had nine nine different... So while the focus was really on phone companies, there were also big implications for the future of how we would connect to the internet.
[00:17:11] What the 96 act did is that it created the opportunity for what were called CLX.
[00:18:51] the new CLX, these new competitive local exchange carriers, some of them began to offer DSL service//which was a slightly higher speed way to access the Internet, because prior to about the mid 90s, the only real way you could access the Internet was through a very slow dial up telephone modem.
[00:19:48] that really led to the birth of the broadband Internet. And once we had higher speeds on the Internet, well, then we could have things like video, things like real time communications
But it also created some issues for the connection points. Who should have access to them? [00:21:49] the 96 act also facilitated and kind of laid the groundwork for things like more exchange points that could be run by private kind of companies when a company is not affiliated with the existing telephone companies// And so really, the 96 act kind of paved the way for the broadband Internet, but it also paved the way for an interconnected Internet where.
[00:22:32] We're any company with a relatively low investment could come in to a really a neutral place to exchange traffic. And why would it even matter if there were neutral exchange points anyway?
[00:22:44] Part of the problem following the 96 act is that actually the incumbent phone companies, you know, were were forced to allow other companies to connect into their central offices. But they were very, very resistant in terms of permitting equal access to the central offices. And so the competitive companies had to jump through all kinds of difficult bureaucratic hoops just to get access to their own equipment in the central offices. And that really was one of the sparks that led to kind of a proliferation of exchange points, places where traffic could be exchanged outside of a phone company facilities. Sothe 96 act lay a foundation for kind of all of the physical part of many of the physical parts of the Internet that that that really underlie it today
So the internet is gaining steam.... Commercial companies could get involved. Like train tracks or phone lines, there was a huge opportunity for companies to come in and make money.... owning connections points - controlling the interconnection that was growing every day. Those commercial companies increasingly wanted control over the wires, and the decentralized nature of the internet - the thing that made it so special and resilient - was at risk.
[00:25:07] cable companies, which like the phone companies, had wires running all around the nation. The cable companies said, well, hey, we can get in and try to compete providing communications over our wires.
[00:25:24] But then there was a pretty big debate. Would the cable companies and networks be kind of controlled by a single company that only allowed one company to provide Internet access over the cable network? Or would the cable companies be required to have an open access regime where any number of companies, small CLX and others could compete to provide services to cable companies just like CLX could compete to provide services to phone customers? Could could small companies provide services over the cable?
[00:26:46] And at the end of the day, cable companies really prevailed. They they were not forced to allow competitive access over over their networks. This eventually leads us to the net neutrality debate which relates to open access. Should phone companies or cable companies pick and choose preferred internet providers? Should they get to control what consumers have access to?
[00:27:59] You know, could if I signed up for four AT&T Internet, would AT&T be able to say that, that the only search engine I could use would be Bing? You know, I couldn't use Google or would I have a choice to go to any search engine I want? And, you know, at least it's been a struggle on the network net neutrality arena. In the end, the Telecommunications Act - where regulation around the internet was largely a footnote - actually had surprising, long lasting impacts on the internet...and that includes infrastructure.
[00:35:39] the 96 act really did pave the way for all of the pieces of infrastructure that make up the modern Internet today. So exchange points and and non phone company carriers of backbone traffic and and on//00:37:28And so there are a whole range of critical companies that are kind of behind the scenes. And and those companies were really facilitated. Their existence was facilitated by the 96 act. To this day, the ‘96 Act has helped create a system where the internet could get faster and faster - BECAUSE it was decentralized.
[00:32:07] what's also developed because of Internet connection is services both run by network operators, but also run by other companies that that allow that video to be hosted in dozens or hundreds of places around the Internet so that if I want to connect to it, I only have to connect to something that's much, much closer than California. I connect to something that's just, you know, 10, 20 miles away. And because there's good connections, it only goes through two or three connections or what we call two or three hops on the Internet to get to me from the source to the to the recipient. And so, I mean, it was critically important that that we. You know, kind of enable very broad, ubiquitous connection to allow, you know, large and small speakers to be able to reach their audiences quickly and reliably and without necessarily being throttled or regulated by an individual network provider
But with regulation comes great responsibility - and the commercial wave of growth that followed meant more possible breaking points for the internet. More in a minute.
Act 4: The Growth of the Net as We Know It
It looks like a totally nondescript building.
Equinix is the largest data center company and interconnection business in the entire world. This is a business that was built ultimately to help the Internet scale and reach its potential.
That’s Jay Adelson - founder of Equinix, and he’s the guy behind these data centers. He’s someone who witnessed the growth of the commercial side of the internet first hand. For him, the 90s was when the internet really stopped being nice... and started getting REAL. Yes, that was a Real World reference. Your welcome.
And at that time, there really wasn't an understanding, both with the government or even the research community, that the Internet would be this commercial amazing thing that it's become. And so the idea at the time was to.
[00:06:02] Basically hand the reigns of the Internet off to the top five largest telephone companies at the time. Companies like AT&T and Sprint.
[00:06:12] This brings us back to a term we’ve talked about before...“network access points” - basically exchange points. And if you wanted to connect to anyone else, you would have to pay them to connect to them. Now, this probably made sense at the time because telephone companies really they weren't officially in the Internet business. But as soon as the Internet started to get exciting in 95 and 96, they started selling Internet services. And so ultimately, the people who controlled the airports, it would be like an airport was owned by an airline and then charged attacks to their competitors to come land at their airport. This was still early days - we’re talking very few browsers, very few websites and a handful of search engines like Netscape...and all those old-school message boards where you could find people discussing just about everything...people could see the potential but we weren’t there yet. Some classic methods of expanding infrastructure served as a blueprint, but only to a point:
[00:16:18] railroads were instrumental in how the Internet got actually scaled because they provided the right of way to run fiber optic cable or copper cable without interruption over long distances.
And in fact, when we talk about the initial development of the telephone industry, the telecom industry, it was really the railroads that provided the, you know, the telegraph and the telephone connections that backbone that would haul traffic from one region to another region, you know, much like the interstate highway system, had to keep growing and it got expensive. And so I think it was a logical conclusion that a single network wasn't going to scale economically. But then there was a secondary question which was, How do we handle the technology, the routing, the you know, you know, all of these autonomous networks, how do we prioritize traffic across this network? ] A big piece of that would be nodes - places where different networks could connect with each other. The internet would need to be built differently than centralized systems, and it would be set apart from previous ways information could travel - whether over phone lines or train tracks. The nodes or exchange points where different networks met, became a bottleneck.
[00:35:19] right around 1997, the operators of the Internet, the content, the providers who who created content on the Internet and the enterprises that started to use it for critical services became concerned that the central exchange points were both congested and overpriced and the performance of the Internet was never going to get any better// It was starting to decline pretty quickly. And so all of these business models were basically threatened by this by these problems at these core exchanges, so the concept of a neutral exchange model was a potential solution to this problem. A technological innovation was needed - and Jay Adelson had an idea.
[00:23:36] The truth is, is that a real simple solution is sometimes the best and much like, you know, telephone operators plugging cables into, you know, different patch panels, we were doing that in a data center] Adelson’s Equinix was a company built on a philosophy....born out of the early internet culture that believed in open access. More equality would help the internet grow. And that philosophy also became a business model.
[00:24:27] What we did is we said, let's make a flat rated cross connect the business model or a flat rated port, charge the business model and not be a network service provider ourselves. And that innovation, that concept, which may seem obvious, was not obvious]
VAN CAMP [00:54:00]
certainly at cocktail parties, they think I'm either a health club or a credit bureau Peter Van Camp is executive Chairman at Equinix. He joined the founders - Jay Adelson and Andy Smith - very early on at the company. And he jokes that even all these years later, people don’t always know which companies are responsible for those lightning fast internet connections.
Even if you have access to wires and towers, you still need many networks to be able to interact freely for the internet to continue to scale - and businesses to be built. But to build the right kind of neutral infrastructure needed.
[00:07:40] I don't think Equinix could have been created if it wasn't for the time we were in. This is the late 90s, if you remember and you remember a dotcom bubble formed around the Internet opportunity that everybody was staring at capital flowing so freely towards Internet opportunities enabled this to happen. But even the earliest days of Equinix required a great deal of capital.
The original plan was to cover six major markets around the US: New York, Washington, D.C., which is really Northern Virginia. It was Dallas, Chicago, Silicon Valley, Los Angeles. Then a bigger ecosystem could emerge around these neutral connection points.
[00:09:01] most data centers at the time, most of them actually were owned by the the Internet networks or the telecommunications companies who were offering Internet access. And those were use originally created to have more value for the customers of those networking companies by being able to host the websites or maybe IT platforms what have you if you're just outsourcing, sourcing your data center requirement. //But when they recognized the growth of the Internet and the ability to scale it and be global in its reach, it was very difficult from an AT&T data center to then stretch the fiber and extend the network over to a sprint data center. // all of these major backbones realized that there would be value to interconnecting at that point in time, but they weren't going to trust going to each other's data center. And then the cost of this intricate, almost spaghetti like fab fabric that was going to in each data center just was not going to make sense from a cost or scaling standpoint] Equinix extending a neutral playing field to all these companies was vital in increasing interconnection - helping data centers, networks, and a growing number of companies “talk” to each other efficiently.
[00:12:31] the reality of how fast it needed to grow and what it needed to be from a just a reliability and resilience standpoint, it needed just infrastructure that was specifically designed for this and was was run to support that that growth. //they created this vision, the goal was to be the stewards of the Internet's infrastructure. The dot com boom in the late 90s helped the company begin, and grow. But the internet bust put Equinix in danger as well.
[00:25:31] I think at one point in time we had in our pipeline five different online pet stores. Really, the world needs an online pet store and not only one, but five.
[00:25:44] And so this was the vision that anybody with a hard drive and a great idea was going to put up a site to be able to start to market that idea early// it was that commercial opportunity that brought a lot of new ideas to the Internet
[00:18:21] And at the end of the decade, all of this growth that was projected would eventually come, but just not at the pace that all the analysts were guessing, not at the pace that all the investments were being made. // as the the bubble began to leak, everyone was seeing, oh, my, this is not going to end well// And Equinix, in fact, had a number of customers that we had acquired and a number of cages to use that term of what's inside a data center dedicated to a specific customer in a multi datacenter environment. A number of our cages never were filled.// And so we lost many customers, turned a great deal of committed revenue through the bankruptcies that took place
Act V: Infrastructure and Interconnection, Then and Now]
Though the recovery was slow and painful, the internet didn’t take long to right itself. And as the internet grew and changed, the needs of infrastructure also changed over time. And that in turn changed what customers could have - and what they might want.
In in the 2000s There was a change of philosophy from one of constraint where people thought, well, there's no way that I'm ever going to be able to to deliver enough. Data across a phone line to actually do this type of application remotely, that shifted really quickly. It shifted because largely on part of due to the neutral exchanges, the cost of traffic dropped precipitously and went from hundreds of dollars per megabit or thousands of dollars per megabit to pennies More giant data centers were being built - places that held interconnection points but also a lot of processing power.
[00:46:31] And when that happened through the competition and the neutrality and the and the availability of bandwidth, all of that allowed these new models around data centers to start to emerge where where instead of, you know, a company having their computer room in their office// it moved to one where it actually made financial sense to take all of that content, whether it's your email or your files or video, and move it to the same building that the interconnection was at
[00:49:33] So like a year and a half. Yeah. Of my daughter's childhood went from 20000 square foot data centers to 500000 square foot data centers. And that that meant that. Not only would Equinix have to build larger data centers to handle hundreds or thousands of different types of customers coming in there, but also the very largest of all of the content providers, companies like Google and Facebook, would need to really start building their own because they were just they would take 100000 square feet in one in one city. And this is in the early 2000s. Now, if you look at it, fast forward to to 2020, a company like Amazon or Google has hundreds of data centers all over the world and they are enormous and single purpose.
As the ecosystem evolved, the tech giants like Amazon, Microsoft and Google continued to build out their own massive data centers. But neutral exchange points - the place where interconnection happens - have continued to be vital. And the health of that ecosystem - that solid, dependable infrastructure is built - and the fact that it’s interconnected - has proven to be vital as more challenges developed over time. Like the breaking point that came in 2020, thanks to the global covid-19 pandemic. So we had the data centers, the cloud....and still more was needed. All of this left company leadership thinking about physical infrastructure again- and how to make sure there was enough. Again, Peter Van Camp.
VAN CAMP [00:31:06] you know, digital transformation for any company is a board room
issue now. And so any company from a grocery store to to hardware, they're a digital and technology company now in the way they're going to need to operate and obviously in a pandemic time like this
[00:49:21] We were just overwhelmed. We had to manage our own curve to use that medical term of what was going on in the ICU, use of hospitals all over America. But all of our customers saw this digital wave of need coming at them and as a result, they were all bolstering their infrastructure, adding to the scale that would be ready to absorb this growth. And so we were literally booking appointments in data centers for customers to be able to come in and upgrade and certainly just to participate Whether we’re talking about the dot com boom of the 90s or the zoom boom of 2020, Jay Adelson, founder of Equinix, reiterates just how important a commitment to open access really is to the internet today - and in the future.
I think everybody is still very committed to to an open framework by which they can interconnect with each other and do what they need to do within their sort of their various camps. I don't see a technology breaking point coming anytime soon.
[00:53:10] I do believe that the neutral exchange model, combined with edge computing and moving applications to where they need to be, it scales pretty well. It's expensive, but it scales pretty well] But even without a technology breaking point, there are still other things that could threaten future growth.
[00:53:26] I think the problems that we're facing is sort of the the Pandora's box we open by making it so easy and so globalized for people to communicate so freely, everything that I dreamed of. The Internet could be for good, has now been realized to be used for evil, and I hate to say it like that, I don't think that there's a Dr. Evil sitting behind, you know, petting a cat. But what we've seen in terms of the abuse of people's data, the political climate misinformation, the culture we've built around using the Internet, that's the breaking point that we're facing right now. And and neutral exchange and interconnection, while I do believe, you know, historically was necessary and I don't regret creating that scalability. I think that now we have to be sensitive, you know, in net neutrality debates and all of these things are related to this, we have to be sensitive to this problem of information and how it's used. And and I and I do believe that it's it's not just the responsibility of the consumers. And I don't believe it's just the responsibility of the Facebook's and the Googles either. I think that it's a shared responsibility
Join us throughout the season as we break down different layers of the internet’s infrastructure, to tell you hidden stories and history you’ve never heard. Next time on Traceroute.
Thanks for listening to Traceroute, an Equinix Metal production. Our theme music is by Ty Gibbons. Be sure to subscribe in Apple, Spotify, or wherever you get your podcasts.