How the Internet Works

Every time you load a web page, data makes a journey that spans continents, passes through dozens of machines, and arrives in milliseconds. It feels instant, but a lot is happening under the hood. This lesson explains the mechanics — the cables, the packets, the routers, and the handshakes that make it all work.

The Physical Layer

The internet is, at its core, physical infrastructure. Fiber optic cables run under oceans connecting continents. Copper wire and coaxial cable run into homes. Cell towers and Wi-Fi routers move data wirelessly over radio frequencies. Data travels as pulses of light through fiber or as electrical signals through copper — the medium changes, but the underlying principle is the same: information moves as a physical signal.

Undersea cables are particularly important. When you load a website hosted in the US from Europe, your data almost certainly travels through one of these cables — systems like FLAG, SEA-ME-WE, or the newer systems laid by companies like Google and Meta. There are hundreds of such cables forming the backbone of global internet connectivity.

Packets

Data on the internet doesn't travel as a single continuous stream. It's broken into small chunks called packets — typically 1,500 bytes or less. Each packet contains the actual data payload, plus metadata: the source IP, destination IP, and sequence information so the packets can be reassembled in the right order.

This design is deliberate. Breaking data into packets means multiple devices can share the same network link at the same time — your video stream and your colleague's file download coexist on the same wire. It also makes the network resilient: if one packet is lost or takes a different route, only that packet needs to be retransmitted, not the entire transfer.

IP Addresses

Every device connected to the internet has an IP address — a numerical label that acts like a postal address. When you send data somewhere, the packets are stamped with your IP (the source) and the destination's IP. Every router along the path uses these addresses to figure out where to forward the packet next.

There are two versions in common use: IPv4 (e.g., 93.184.216.34) and IPv6 (e.g., 2606:2800:220:1:248:1893:25c8:1946). More on those in the next lesson.

Routers

Routers are the traffic directors of the internet. Each router knows its neighboring routers and, using a routing table, makes a local decision about where to forward each packet. No single router knows the full path to every destination — it just knows enough to pass the packet one hop closer.

This hop-by-hop forwarding is how packets travel across the world. You can see it in action with traceroute:

traceroute google.com

traceroute to google.com (142.250.185.78), 30 hops max, 60 byte packets
 1  192.168.1.1       0.5 ms    # your home router
 2  10.0.0.1          3.1 ms    # ISP gateway
 3  72.14.215.212     8.2 ms    # ISP backbone
 4  209.85.175.100   12.4 ms    # Google's network
 5  142.250.185.78   13.1 ms    # destination

Each line is a router the packet passed through. The latency numbers show how long it took to reach each hop. Most transatlantic connections involve 10–20 hops.

ISPs

You don't connect directly to the internet — you connect to an Internet Service Provider, which connects you to the internet. Your home router establishes a connection to your ISP's local point of presence.

ISPs aren't all equal. At the top are Tier 1 ISPs — companies like Lumen, AT&T, Telia, and NTT. These operators own or have rights to the physical infrastructure connecting continents. They peer with each other under settlement-free agreements, meaning they exchange traffic without paying each other. Below them are Tier 2 ISPs who pay Tier 1s for transit, and Tier 3 ISPs (your home internet provider) who pay Tier 2s.

Your packet climbing from your ISP up to a Tier 1 and then back down to another ISP's network to reach its destination is a standard journey.

The Full Journey

Let's trace what happens when you type google.com and hit enter.

1. DNS resolution — Your browser needs to convert google.com to an IP address. It checks its local cache, then asks your OS, then queries your ISP's DNS resolver, which ultimately finds the answer and returns it. (You'll learn how DNS works in detail later in this series.)
2. TCP connection — Your browser opens a TCP connection to Google's server IP. This is a three-way handshake: SYN → SYN-ACK → ACK. After this, a reliable channel exists between your browser and Google's server.
3. TLS handshake — For HTTPS, your browser and the server negotiate encryption, verify Google's certificate, and establish a secure session.
4. HTTP request — Your browser sends GET / HTTP/2 with headers (cookies, accepted content types, etc.) over the encrypted connection.
5. Packets travel — The request is broken into packets. Each travels hop-by-hop through your router, your ISP, backbone infrastructure, and into Google's network.
6. Server response — Google's server processes the request and sends back HTML, CSS, JavaScript, and other resources — all broken into packets.
7. Reassembly and render — Packets arrive (possibly out of order) and are reassembled by TCP. Your browser receives the response and renders the page.

The entire round trip typically takes 20–80ms for nearby servers, or 150–300ms for servers on the other side of the world.

Latency vs Bandwidth

These two terms describe different things and matter in different situations.

Bandwidth is how much data can move per second — think of it as the width of a highway. A 1 Gbps fiber connection can transfer a gigabit every second.

Latency is how long a single packet takes to travel from source to destination — think of it as the distance between two cities. Even with infinite bandwidth, the speed of light sets a floor on latency. New York to London is ~70ms minimum (the light-speed limit for that distance).

Latency matters most for interactive applications: real-time games, video calls, trading systems. Bandwidth matters most for large transfers: video streaming, file downloads. A connection with high bandwidth but high latency will still feel sluggish for anything interactive.

Real-time applications like video calls are highly sensitive to latency. A 300ms round trip produces noticeable delay in conversation. A 50ms round trip feels instant. This is why companies like Netflix don't serve video from a single origin server — they use CDNs.

Netflix operates one of the largest CDNs in the world, called Open Connect. Netflix boxes sit inside ISP data centers in hundreds of cities. When you press play, your traffic goes to a nearby cache, not to Netflix's origin in Virginia. This keeps the data local, avoids ISP peering bottlenecks, and delivers consistent video quality regardless of what's happening on the broader internet.

Key Takeaways

• The internet is physical cables, radio signals, and fiber — data travels as light and electrical pulses
• Data is broken into packets that travel independently and are reassembled at the destination
• Routers make hop-by-hop forwarding decisions — no single router knows the full path
• ISPs connect end users to the internet; Tier 1 ISPs form the backbone that ties everything together
• Bandwidth measures capacity (bits per second); latency measures travel time (milliseconds)
• Modern applications like Netflix use CDNs to bring content physically closer to users

Next up: how devices are addressed on the internet and how large networks are subdivided — IP Addresses and Subnetting.