From NSF to ISP

  |    
技術の歴史

Command Line Heroes • • From NSF to ISP | Command Line Heroes

From NSF to ISP | Command Line Heroes

About the episode

1995 was the year that ISPs became the dominant gateway to the information superhighway. But how’d we go from ARPANET all the way to that? It turns out, none of it would have happened without a team of intrepid engineers at the University of Michigan.

Marc Weber tells us how a tension between academics and the military set the next evolution of the ARPANET. Douglas Van Houweling discusses the work his MERIT team did at the University of Michigan to build the national backbone of the NSFNET. Elise Gerich, MERIT’s systems manager, talks about how they made the leap from a T1 connection to a T3 to handle traffic from their growing network. And Janet Abbate emphasizes how all this set the stage for the commercialized internet that birthed the dot-com boom in 1995.

Command Line Heroes Team Red Hat original show

購読する

Subscribe here:

Listen on Apple Podcasts Listen on Spotify Subscribe via RSS Feed

トランスクリプト

In the spring of 1995, disruption was in the air. A quiet and invisible disruption that few would have noticed. A program called the NSFNET was shutting down. It didn't sound like much, did it? But as those switches were thrown and a bank of servers grew cold, a part of America was remade. You've maybe heard of the ARPANET, sort of the godfather of the internet, right? An early set of connections between a few military sites and universities. And I know you're familiar with today's internet, but there's a missing link between those two points, another network that thrived for a while in the 80s and 90s. This missing link connects the early days of publicly funded innovation and the later days of commercialized networks. It was called the NSFNET, the National Science Foundation Network, and it served as the internet's backbone for years. The little known story of its origin, and eventual demise in 1995, reveals a lot about how we got the online life we have today. It's the story of how the internet was privatized. I'm Saron Yitbarek, and this is Command Line Heroes, an original podcast from Red Hat. All season we're exploring one fateful year, 1995, to understand how today's tech reality was formed in that single moment. We're learning how one year became a catalyst for our whole future. And no moment in 1995 tells you more about that massive change than April 30th, when the Internet's backbone, the NSFNET, was switched off. The story of that fateful moment begins, though, back in the early 1980s. The United States was falling behind other countries in the field of large-scale computing, and the National Science Foundation, the NSF, decided to remedy that by creating supercomputing centers all around the states. Only problem was scientists needed access to those supercomputers. The logical thing to do was simply ring up the Defense Department and ask them to upgrade the network that already existed, the ARPANET. And that's exactly what the NSF did. They made a deal, paid the Defense Department $2 million for their trouble and waited. And what the NSF might not have considered was that the Defense Department wasn't so excited about funneling more traffic to the ARPANET. Here's Marc Weber, the curatorial director of the Internet History Program at the Computer History Museum. The chance of people getting into something that's relatively open and not properly secured was going up. Try and look at this from the perspective of the Defense Communication Agency. Sure, it's nice that all these academics get to share information across the ARPANET, but that made state secrets vulnerable in the process. They grew anxious about the security of the network. Back then in 1983, there was a sci-fi movie called War Games. A teenage Matthew Broderick finds a back door into the U.S. Military's computer system, and thinking he's playing a video game, he almost starts World War III. Sure, that sounds far-fetched, but that movie reflected real concerns. The military in the 1980s was growing frustrated that the ARPANET was open to outsiders. Researchers tended to want openness and to share material without worrying too much about copyright or permission. And the military obviously wants security and traceability. So there was that tension from the start. There was a point they even thought about just getting rid of the ARPANET. They might have gone off in completely their own direction. Eventually the scientists over at NSF got the message. The Defense Department was clearly not enthusiastic about this. That's Douglas Van Houweling, a professor at the University of Michigan and a pioneer from the early days of the internet. He told us the Defense Department never did get around to implementing those connections that the NSF needed. As a result, the National Science Foundation realized it had to take a different course. In fact, they planned to build a network all their own. In 1985, the NSFNET was born. Independent of the military, the National Science Foundation could provide academics with email, with file transfer protocols so they can upload and download files, and even remote access to time-sharing systems. So they were offering a lot of what the ARPANET had offered. But no, that's not the end of the story. The NSFNET had its heart in the right place, and its protocols in the right place too, but they needed help with the technical implementation. They needed someone who could build this thing for them. And luckily, they found exactly the right team for the job. I was working at the University of Michigan, and we had created a consortium of universities in Michigan called MERIT. MERIT stood for Michigan Educational Research Information Triad. It was a consortium that Van Houweling and his colleagues had put together to share computing resources among Michigan's universities. And when the NSF came looking for a contractor to build their network, MERIT seemed like a perfect fit. We had already demonstrated that we could do this kind of thing on a state level. We had the expertise, we had the relationships with the universities, and we had the vision for what this could become. In 1987, MERIT won the contract to build and operate the NSFNET backbone. They would connect five supercomputing centers: San Diego, Colorado, Illinois, Pittsburgh, and Cornell. But this wasn't just about connecting a few supercomputers. Van Houweling and his team had a much bigger vision. We realized that this wasn't just about supercomputing. This was about creating a national research network that could connect universities and research institutions across the country. It was about enabling collaboration and resource sharing on a scale that had never been possible before. The NSFNET started small, but it grew fast. Really fast. And that growth was thanks in large part to Elise Gerich, who was MERIT's systems manager. When we first started, we had a T1 line, which was 1.5 megabits per second. That sounds tiny today, but back then it was a big deal. We were connecting all these universities and research institutions, and the traffic just kept growing and growing. Elise Gerich became known as the "Mother of the Internet" for her work managing the NSFNET. She watched the network grow from connecting a handful of supercomputers to connecting thousands of institutions across the country. By 1989, we realized that the T1 line wasn't going to be enough. The traffic was overwhelming our capacity. We needed to upgrade to a T3 line, which was 45 megabits per second. That was a huge jump, but we could see that we were going to need it. The upgrade to T3 was a massive undertaking. It required new equipment, new software, and new protocols. But it also represented something bigger: the realization that the internet was becoming something much larger than anyone had originally envisioned. We weren't just connecting researchers anymore. We were connecting whole universities, and those universities were connecting their students, their staff, their libraries. The network was becoming a platform for all kinds of communication and collaboration. As the NSFNET grew, it began to attract commercial interest. Companies wanted to use the network for business purposes, but there was one problem: the NSF had an "Acceptable Use Policy" that prohibited commercial activity on the network. The Acceptable Use Policy was designed to ensure that the network was used for education and research, not for commercial purposes. But as the network grew, the line between education and commerce became increasingly blurred. The tension between the network's educational mission and commercial pressures was growing. Companies were creating their own networks to get around the restrictions, and there was increasing pressure from Congress and the business community to open up the internet to commercial use. We knew that eventually, the commercial sector was going to have to take over. The network was getting too big and too expensive for the government to fund indefinitely. And there were legitimate questions about whether the government should be in the business of running a communications network. By the early 1990s, the writing was on the wall. The NSF began planning for the privatization of the internet. They would shut down the NSFNET backbone and hand over responsibility to commercial Internet Service Providers, or ISPs. The transition was incredibly complex. We had hundreds of networks connected to the NSFNET, and each one of them needed to find a new way to connect to the internet. We spent years planning for this transition, making sure that everyone would be able to stay connected. The transition wasn't just technical—it was also cultural. The internet was moving from a world of shared resources and collaborative research to a world of competition and commerce. It was bittersweet, in a way. We had built this amazing network that was enabling incredible collaboration and innovation. But we also knew that privatization was necessary for the internet to reach its full potential. As 1995 approached, the pressure was mounting. The NSF had announced that the NSFNET would be shut down on April 30th, 1995. Everyone connected to the network needed to find alternative arrangements. The last few months were incredibly stressful. We were getting calls from network administrators all over the country, asking for help with the transition. Some organizations were ready, but others were scrambling at the last minute. Despite the stress and uncertainty, there was also excitement about what was coming next. The privatization of the internet would unleash new possibilities for innovation and growth. We knew that once the commercial sector took over, the internet would grow faster and reach more people than ever before. We just hoped that it would retain some of the collaborative spirit that had made it so successful in the first place. The day finally arrived: April 30th, 1995. At the University of Michigan, Elise Gerich and her team prepared to shut down the NSFNET for the last time. It was surreal. We had spent so many years building and maintaining this network, and now we were going to turn it off forever. But we also knew that we were making history. We were ushering in a new era of the internet. One by one, they shut down the exterior nodal switching subsystems that had been the backbone of the internet for nearly a decade. As each system went offline, a piece of internet history came to an end. When we shut down the last system, there was a moment of silence. We all knew that something important had just ended. But we also knew that something even more important was just beginning. With the NSFNET gone, the internet was now fully in the hands of commercial ISPs. Companies like PSINet, UUNet, and BBN were ready to take over, providing internet access to businesses and consumers for the first time. The transformation was immediate and dramatic. Within months, we saw an explosion of commercial activity on the internet. Companies that had been waiting for years to get online finally had their chance. The dot-com boom was about to begin. But it was built on the foundation that Van Houweling, Gerich, and their colleagues had laid with the NSFNET. The technical infrastructure, the networking protocols, the culture of openness and collaboration—all of these were legacies of the academic internet. I think people sometimes forget that the commercial internet didn't spring up out of nowhere. It was built on decades of research and development funded by the government and carried out by universities. The NSFNET was a crucial bridge between the early ARPANET and the modern internet. The impact of the NSFNET's closure went far beyond just changing how people connected to the internet. It fundamentally altered the nature of online communication and commerce. Once commercial activity was allowed, the internet became a platform for all kinds of business innovation. E-commerce, online banking, digital marketing—none of this would have been possible under the old Acceptable Use Policy. But the transition also raised important questions about the future of the internet. Who would be responsible for maintaining and upgrading the infrastructure? How would access be ensured for universities and other non-commercial users? The privatization of the internet was presented as a solution to the problem of funding and scalability. But it also created new problems and inequalities that we're still dealing with today. Janet Abbate is a professor of science, technology, and society at Virginia Tech. She's studied the history of the internet and its transformation from a government-funded research network to a commercial platform. When the internet was run by the government and universities, there was a commitment to universal access and open sharing of information. When it was privatized, that commitment was replaced by market forces and profit motives. This shift had profound implications for how the internet developed over the following decades. The focus shifted from research and education to entertainment and commerce. We gained a lot from privatization—faster speeds, wider access, more services. But we also lost something important: the sense that the internet was a public good, a shared resource that belonged to everyone. Despite these concerns, the privatization of the internet was largely seen as a success. The network grew exponentially, reaching millions of new users and enabling new forms of communication and commerce. The growth was astounding. In just a few years, the internet went from connecting a few thousand researchers to connecting millions of people around the world. That kind of growth would never have been possible with government funding alone. But even as they celebrated the success of privatization, the pioneers of the NSFNET remained committed to the values that had guided them from the beginning. We always believed that the internet should be open and accessible to everyone. Even as it became commercialized, we hoped that it would retain some of that original spirit of collaboration and sharing. Today, as we grapple with issues like net neutrality, digital privacy, and platform monopolies, the legacy of the NSFNET remains relevant. The tensions between public and private control of the internet are still very much alive. I think we need to remember that the internet wasn't inevitable. The choices that were made in the 1980s and 1990s—about funding, about governance, about access—shaped the internet we have today. And we can still make different choices going forward. The story of the NSFNET reminds us that the internet we know today is the result of countless decisions made by engineers, policymakers, and visionaries over many decades. I'm proud of what we accomplished with the NSFNET. We helped create the foundation for the modern internet. But I also think it's important to remember that the internet is still evolving, and we all have a role to play in shaping its future. As we look back at April 30th, 1995, we can see it as both an ending and a beginning. The end of the government-funded internet, and the beginning of the commercial internet that we know today. But it wasn't until February that basically MERIT sent the termination notice of the service to all of its 19 locations, and that's when people became desperate. They started sending us so many configuration changes. Some of the regionals, at the last month, said, "Well, we're not going to make it; our connection from our new service provider won't be here." And it's like, oh my gosh, what are we going to do now? We're shutting stuff down. A couple of regionals said, "Well, we just don't want to do it." And we said, "Sorry, this is going to happen." It was going to happen and it was going to change everything. Yes, there had been dial-up services, like America Online, which paid for their own infrastructure. Banks and airlines had their own networks too. But after April 30, 1995, vastly more businesses and entrepreneurs were going to be coming to the table. On the NSFNET's very last day, a dozen staff gathered at the University of Michigan to finally shut it down forever. One by one, they shut off the ENSS's, the exterior nodal switching subsystems. And that evening, the NSFNET breathed its final breath. For some, that day in 1995 came to be known as Internet Independence Day, the day the internet truly left the government's domain. It launched a new age of e-commerce and social media, a whole new landscape of online experience. But the internet's history is never quite that simple. As Janet Abbate notes, the innovations that the academic world brings to our network world did not stop that day. The National Science Foundation never stopped funding research in high-performance computing and networking. I think their current focus is on wireless networking, among other things, funding advanced cutting edge networks for universities and scientists. From their point of view, the NSFNET was an early episode in an ongoing research enterprise. That research enterprise lives on. What is sometimes missing today though, is the oversight—the policy that government can provide. Well, in a way I'm not sure what we have today is any more sustainable than the NSFNET was in 1995. We have conflicts we don't exactly know how to resolve. And we didn't cure the problem in 1995 by throwing it to the private sector. I think there was this hope that the private sector is kind of a universal solvent and everything will be fixed by it, and that didn't really happen. So I think we have to face up to the fact that there's always going to be conflicts on the internet, and simply hoping people will exercise self-restraint is not really a substitute for real policy mechanisms. In recent years, an open internet order reasserted government control and lobbyists for private companies pushed back. That balance between public and private is something we can lose sight of when we roam the internet today. Sometimes it feels like one big commercial, but that's just not the whole picture. Here's Douglas Van Houweling one last time to remind us of the internet's roots. I think what a lot of people don't understand is that the internet today stemmed from the need of the research community, first in the United States, and then all over the world, to be able to do their science. The NSFNET was built to support researchers across the nation in working with one another, in accessing the supercomputers, in accessing data that they needed to do their research. So as hard as it may be to realize today, the motivation for NSFNET and the motivation for the first large-scale generation of the internet had nothing to do with making any money. It was absolutely about advancing knowledge and supporting the research community. However the internet evolves from here, it's most exciting advances won't just come from some new startup. Fundamental changes still tend to come from that place where private innovation meets the pure science and research that underpins our online lives. That was the marriage of public and private that had pushed the NSFNET to a point of such fantastic possibilities by 1995. For a brief moment in between the ARPANET and today's internet, the NSFNET shined a light on what's possible when government, academics, corporations, all the players, start playing on one team. The story of the young internet sometimes gets told as a journey from the public realm to the private realm, but that's not exactly how I see it. Our internet today is a private realm built on top of a public realm, an online experience that relies on decades of public and academic work. We should remember that when we're looking around for the next big thing, because it might just come from that sweet spot, that moment in 1995 when the NSFNET just couldn't stop growing. Next time it's the web's dominant language, HTML. We're exploring its origins, its standardization. And we're also asking whether English programming languages are ready to embrace a truly global perspective. Until then, I'm Saron Yitbarek, and this is Command Line Heroes, an original podcast from Red Hat. Keep on codin'.

More about 技術の歴史

未来を想像するには、過去を振り返って、感動と刺激を与えるイノベーションの事例を発見することが重要です。

ブログ投稿

Red Hat Insights が Red Hat Lightspeed へと進化：AI を活用した管理を加速

2025年 11月 4日  |  Brent Midwood

...

Automation and management

ブログ投稿

Discover が 2 日間で AWS の年間予算から 140 万ドルを削減した方法

2025年 10月 21日  |  Alex Handy

...

Cloud services

About the show

Command Line Heroes

During its run from 2018 to 2022, Command Line Heroes shared the epic true stories of developers, programmers, hackers, geeks, and open source rebels, and how they revolutionized the technology landscape. Relive our journey through tech history, and use #CommandLinePod to share your favorite episodes.