Tubes

Tubes Plot Summary

Introduction

In the winter of 2008, author Andrew Blum's internet connection faltered. A technician traced the problem to a backyard cable chewed by a squirrel. This mundane incident sparked an extraordinary revelation: the internet, this seemingly ethereal entity, was actually a physical thing that could be damaged by rodent teeth. For something so central to our daily lives, we know surprisingly little about where the internet actually exists. The internet is not a mystical cloud floating above us, but rather a tangible infrastructure of cables, data centers, and exchange points spanning the globe. It connects continents through undersea cables, cities through fiber networks, and ultimately, you to the rest of the world. This journey to find the physical internet takes us from the basement of a historic building in Los Angeles where the first internet connection was established, to unmarked buildings in Virginia housing millions of connections, to beaches where undersea cables emerge from the Atlantic. By following the internet's tubes - its physical pathways - we discover not just technological marvels but a profoundly human story about how we've connected ourselves together in the most ambitious communication project ever undertaken.

Chapter 1: The Birth of a Global Network (1969-1989)

On a quiet Saturday afternoon during Labor Day weekend in 1969, a group of computer science graduate students gathered at the University of California, Los Angeles. They were awaiting the delivery of an Interface Message Processor (IMP) - a modified Honeywell DDP-516 minicomputer weighing 900 pounds and costing $80,000. This machine, built by the engineering firm Bolt, Beranek and Newman, would become the first node of ARPANET, the precursor to today's internet. Professor Leonard Kleinrock, who would supervise the installation, had already published groundbreaking work on packet switching - the idea that data could be transmitted efficiently in small chunks rather than continuous streams. This concept, simultaneously developed by Donald Davies in Britain and Paul Baran at RAND Corporation, laid the theoretical foundation for modern networking. While Baran's work aimed to create communications that could survive nuclear attack, the UCLA team was motivated by more peaceful ambitions: enabling computers to share resources and communicate effectively. By the end of October 1969, a second IMP was installed at Stanford Research Institute, and on October 29, the first host-to-host message was transmitted between the two sites. The log book recording this moment, with its simple entry "Talked to SRI host to host," documented what Kleinrock later called "the first breath of the Internet's life." The system soon expanded to include the University of California, Santa Barbara and the University of Utah, forming a four-node network primarily connecting academic institutions. Throughout the 1970s, ARPANET grew steadily but remained a specialized tool used mainly by university computer scientists, defense researchers, and government agencies. The 1980 ARPANET directory was a modest book listing about 5,000 users. Unlike today's sprawling internet, this early network formed a relatively intimate community with shared norms and expectations. Yet within this limited scope, researchers were already discovering the network's potential for collaboration, information sharing, and communication across distances - uses that would later become fundamental to the internet's impact on society.

Chapter 2: From Military Project to Public Infrastructure

The transformation from ARPANET to the modern internet began with a crucial technical evolution on January 1, 1983. After years of planning, all host computers on the network switched from their original protocol (NCP) to TCP/IP (Transmission Control Protocol/Internet Protocol). This wasn't merely an administrative change but represented a fundamental shift in thinking about network architecture. TCP/IP, developed by Vint Cerf and Bob Kahn, established the internet's defining characteristic: a network of networks. Rather than requiring all computers to speak one language, TCP/IP provided a common protocol for different networks to communicate. This design created a decentralized system with no central control point - a principle that remains central to the internet's operation today. As legal scholar Tim Wu noted, this established "a system of tolerated difference" that recognized the autonomy of the network's members. By the mid-1980s, specialized networks were proliferating across academia, government, and business. High-energy physicists had HEPnet, space researchers used SPAN, and magnetic fusion researchers connected through MFENET. European networks like EUnet and EARN emerged alongside regional American networks with names like BARRNet, MIDnet, and NorthWestNet. Yet these networks operated as isolated islands, unable to communicate with each other. The establishment of TCP/IP as a universal language allowed these separate networks to begin connecting, forming what became known as the "Internetwork" or simply, the Internet. From 15 networks in 1982, the number of interconnected networks grew to over 400 by 1986. Computers on these networks increased from 2,000 in 1985 to 159,000 by 1989. The internet was becoming less a technical curiosity and more a vital infrastructure. This period also saw the creation of domain names and the establishment of Network Access Points (NAPs) - centralized hubs where different networks could exchange traffic. These developments transformed the internet from a sprawling theoretical concept into something with specific, important places - nodes that would later become critical infrastructure. By the late 1980s, the internet had moved from being primarily a government research project to becoming a public infrastructure with a clear geography and growing commercial potential.

Chapter 3: The Physical Geography of Digital Exchange

As the 1990s began, the internet's physical infrastructure underwent a dramatic transformation. What had once been confined to university computer labs and government facilities began colonizing new spaces. Telecommunications companies, sensing opportunity in the growing demand for connectivity, started building dedicated fiber-optic networks specifically for internet traffic. Companies like MCI, PSI, UUNet, and Sprint invested billions in laying new cables across countries and continents. This infrastructure didn't develop randomly but followed distinct geographical patterns. The early internet hubs clustered in a few key regions: Silicon Valley and Los Angeles on the West Coast; Boston and Washington, DC, on the East Coast; and international centers in London, Amsterdam, Frankfurt, Tokyo, and later Singapore. These locations weren't chosen arbitrarily - they represented existing centers of technological research, telecommunications infrastructure, and economic activity. Physical geography imposed unavoidable constraints on the supposedly borderless internet. Mountains, oceans, and national boundaries all shaped where cables could be laid. Existing rights-of-way along railroad tracks, highways, and power lines determined the paths of fiber-optic cables. The internet didn't erase geography; rather, it created a new layer atop existing physical and political landscapes, one that both reflected and transformed these underlying realities. By the mid-1990s, when the web browser made the internet accessible to average citizens, there was already a significant physical infrastructure in place. Network engineer Steve Feldman, who worked at Metropolitan Fiber Systems (MFS), helped establish one of the first major internet exchange points called MAE-East (Metropolitan Area Exchange) in Tysons Corner, Virginia. Soon this unassuming location in the suburbs of Washington, DC, was handling half of all internet traffic worldwide. From a physical standpoint, the internet was becoming concentrated in specific, crucial locations. The geography of the early commercial internet reveals an important truth: despite rhetoric about being "everywhere," the internet has always been somewhere. It exists in particular buildings, follows specific routes, and clusters in certain cities and regions. This physical reality contradicts popular metaphors of the internet as a formless cloud. Instead, it is an infrastructure deeply embedded in specific places, shaped by both natural geography and human decisions about where to build.

Chapter 4: Interconnection Points: Where Networks Meet

By the late 1990s, the internet had developed distinct physical nerve centers where networks converged and exchanged traffic. These interconnection points, often called Internet Exchanges (IXs), became the beating heart of the global network. Far from being abstract concepts, they occupied specific buildings and contained tangible equipment that made global connectivity possible. One of the earliest and most significant exchanges was the Palo Alto Internet Exchange (PAIX), established in 1997 in a 1920s-era telephone switching office at 529 Bryant Street. Under the direction of entrepreneur Jay Adelson, PAIX became a neutral meeting ground where competing networks could physically connect their equipment. The building's basement was filled with metal cages housing routers from hundreds of different networks. Above them ran thick bundles of yellow fiber-optic cables connecting these networks together - the physical manifestation of internet exchange. The economic logic of interconnection drove the development of these facilities. Rather than routing traffic through expensive third-party networks, companies found it cheaper and more efficient to directly connect their networks in the same physical location. When two networks wanted to exchange traffic, technicians would literally run a cable from one company's equipment rack to another's. This physical connection, called a "cross-connect," became the fundamental unit of internet interconnection. As the internet grew, similar exchange points appeared globally. Equinix, founded by Adelson after leaving PAIX, built massive interconnection facilities in strategic locations worldwide. One of its most important sites emerged in Ashburn, Virginia, which became known as "Data Center Alley." By the early 2000s, Ashburn housed hundreds of networks in climate-controlled buildings surrounded by security fences and backup power generators. Inside these facilities, thousands of cables connected networks ranging from global telecommunications carriers to major content providers like Google and Facebook. These interconnection facilities became the internet's marketplaces - physical spaces where networks negotiated connections, established relationships, and exchanged traffic. Despite the internet's virtual nature, its most crucial transactions occurred in specific buildings where humans decided how and where to connect their networks. The internet wasn't just mathematical protocols or electronic signals; it was also handshakes, business deals, and yellow cables strung between metal racks in buildings around the world.

Chapter 5: Undersea Cables: The Internet's Hidden Backbone

While internet traffic moves seamlessly across oceans, the physical infrastructure enabling this global connectivity remains largely invisible: undersea fiber-optic cables. These slender tubes, often no thicker than a garden hose, carry over 95% of international communications, including everything from financial transactions to streaming videos to personal emails. The modern undersea cable network began expanding rapidly in the late 1990s and early 2000s, but it follows paths established over 150 years ago for telegraph cables. The first transatlantic telegraph cable was completed in 1858, connecting Europe and North America. Today's internet cables often follow similar routes, landing in the same coastal locations as their Victorian predecessors. Porthcurno, a small cove in Cornwall, England, exemplifies this continuity - it was a major telegraph cable landing site in the 1870s and remains a significant internet cable landing point today. These cables represent remarkable engineering achievements. Each contains strands of glass as thin as human hair, protected by multiple layers of steel wire, copper, and plastic sheathing. Light signals traveling through these fibers carry massive amounts of data - a single modern cable can transmit tens of terabits per second. At approximately 50-mile intervals along the ocean floor, repeaters amplify the light signals using the element erbium, allowing data to travel thousands of miles without degradation. Despite their critical importance, undersea cables remain vulnerable to various threats. Fishing trawlers, ship anchors, earthquakes, and submarine landslides can all damage cables. In 2006, a major earthquake near Taiwan severed seven undersea cables simultaneously, disrupting internet connectivity across East Asia for weeks. Cable repair ships must locate breaks using specialized equipment, retrieve the broken ends from the ocean floor, and splice them back together - a process that can take days or weeks depending on weather conditions and location. The geography of undersea cables reflects both technical constraints and geopolitical realities. Cable routes must avoid underwater mountains, active earthquake zones, and major shipping lanes. They must also navigate political sensitivities, territorial waters, and landing permits. This physical infrastructure creates its own geopolitical tensions, as countries increasingly view control over these vital communication links as matters of national security. Despite the internet's seemingly borderless nature, its physical underpinnings remain subject to the realities of geography, geology, and international relations.

Chapter 6: Data Centers: Where Information Lives

While we experience the internet as immediate and everywhere, our data physically resides in massive warehouse-like structures called data centers. These buildings, often located in nondescript suburban industrial parks or rural areas, house tens of thousands of computer servers storing everything from corporate databases to social media photos to streaming video libraries. They represent the physical manifestation of what marketers call "the cloud." Google's data center in The Dalles, Oregon, illustrates how these facilities operate and why they're located where they are. Built on a 30-acre site beside the Columbia River, this facility consists of multiple warehouse-sized buildings filled with custom-designed servers. Google chose this location for several strategic reasons: abundant hydroelectric power from nearby dams, fiber-optic connections along regional power lines, tax incentives from local government, and a cool climate that reduces cooling costs. These factors - power, connectivity, taxes, and climate - drive data center location decisions worldwide. Inside these facilities, environmental control is paramount. Servers generate enormous heat, requiring sophisticated cooling systems to prevent equipment failure. Power consumption is staggering - a large data center might use as much electricity as a small city. By 2010, data centers globally consumed about 2% of the world's electricity, with consumption growing 12% annually. This energy requirement has led many companies to locate facilities near renewable energy sources like hydroelectric dams, wind farms, or geothermal power plants. Data centers represent different approaches to security and transparency. Google's facility in The Dalles operates with extreme secrecy - visitors are rarely allowed beyond the cafeteria, and employees are instructed not to discuss technical details. In contrast, Facebook's data center in Prineville, Oregon, embraces transparency, offering tours and sharing technical designs through its Open Compute Project. Both facilities, however, maintain strict physical security with multiple checkpoints, biometric authentication, and surveillance systems. The architecture of these buildings reflects their dual nature as both infrastructure and information repositories. Externally, they resemble warehouses or factories - utilitarian structures designed for function rather than form. Internally, they contain raised floors for cooling, backup generators, uninterruptible power supplies, and row upon row of server racks. These physical spaces - with their constant hum of equipment and blinking status lights - form the material foundation of our increasingly digital lives. They remind us that despite metaphors of clouds and virtuality, the internet remains fundamentally physical - a system of machines in buildings connected by cables.

Chapter 7: The Human Element Behind the Internet

Behind the cables, routers, and servers that make up the internet's physical infrastructure are the people who design, build, and maintain these systems. These individuals - network engineers, cable layers, data center technicians, and others - form a global community that keeps the internet functioning 24 hours a day, 365 days a year. The North American Network Operators Group (NANOG) exemplifies this community. Three times yearly, these network engineers gather to discuss technical challenges, share knowledge, and forge the personal relationships that underpin internet operations. Despite managing competing networks, these professionals recognize their interdependence - no single network can function in isolation. At NANOG meetings, engineers from companies like Google, Comcast, Level 3, and hundreds of smaller providers negotiate "peering" arrangements - agreements to directly exchange traffic between their networks. This collaborative ethos extends to the undersea cable world, where engineers like Jol Paling at Global Crossing maintain cables spanning thousands of miles across oceans. When problems arise, these engineers coordinate responses across multiple time zones and international boundaries. In Portugal, when the West Africa Cable System made landfall, a diverse team of Portuguese construction workers, British engineers, and Angolan laborers worked together to connect Africa more fully to the global network. The internet's physical infrastructure also creates local communities. In Prineville, Oregon, Facebook's data center manager Ken Patchett hired locally and integrated the facility into the town's economic and social fabric. In The Dalles, Oregon, Google's data center provides high-tech jobs in a former timber and aluminum manufacturing region. These facilities represent new economic anchors in communities that have lost traditional industries. Most striking is how the internet's physical reality contradicts popular notions of virtuality. Network engineers don't speak of clouds or cyberspace; they talk about specific buildings, cables, and connection points. They understand the internet not as an abstraction but as a tangible infrastructure requiring constant attention and maintenance. As Eric Troyer of Equinix noted, "The cloud is a building. It works like a factory. Bits come in, they get massaged and put together in the right way, then packaged up and sent out." This human element reminds us that the internet isn't a self-sustaining system but a human creation requiring human care. The people who maintain this infrastructure often work in anonymity, their contributions invisible to the billions who rely on their efforts daily. Yet their work forms the hidden foundation of our connected world - a world that exists not in some ethereal realm but in specific places on the map, built and maintained by specific people with specific skills and knowledge.

Summary

The internet, often described as ethereal and placeless, is revealed through this exploration as profoundly physical - a global network of cables, data centers, exchange points, and underwater connections that form a tangible geography. This physical reality contradicts popular metaphors of clouds and virtuality, showing instead a system deeply embedded in specific places and shaped by both natural geography and human decisions. The internet follows existing patterns of development while simultaneously creating new concentrations of connectivity and power in places like Ashburn, Virginia; Palo Alto, California; and the Docklands district of London. What emerges from this journey is the recognition that the internet exists at a fascinating intersection of the physical and virtual worlds. Its infrastructure combines the most advanced technologies with basic elements like yellow cables strung between racks and fiber-optic lines buried along railroad tracks. Understanding this physical dimension helps us better grasp how the internet shapes our lives and societies. It reminds us that despite its apparent immateriality, the internet remains a human creation - built, maintained, and expanded by people making specific choices in specific places. As we increasingly entrust our communications, commerce, and culture to this network, recognizing its physical reality becomes essential to understanding both its possibilities and limitations in our interconnected world.

About Author

Loading...

Andrew Blum

Read more