How Does the Internet Work? Packets, IP Addresses and DNS Explained

Billions of people use the internet daily, yet the mechanics behind a simple page load are often invisible. At its core, the internet is not one network but a mesh of independently owned and operated networks that agree to exchange traffic using common technical rules, according to the Mozilla Developer Network (MDN).

From local cables to a global network

Inside a home or office, computers connect through physical cables (Ethernet) or wireless signals (Wi-Fi) to a device called a network switch, MDN explains. Every device has a unique hardware identifier known as a MAC address, assigned by its manufacturer, and the switch uses these addresses to forward messages to the right machine on that local network.

To link separate local networks together, routers are used instead of switches. Routers rely on IP addresses — numerical labels written as four number groups, such as 142.250.190.78 — rather than MAC addresses. According to MDN, IP addresses are structured so that a router can forward traffic for an entire block of addresses sharing the same prefix, instead of tracking every individual device.

A modem connects a local network to the wider telephone or cable infrastructure, and an Internet Service Provider (ISP) operates the routers that link that connection to other ISP networks. MDN describes the internet itself as simply the sum of all these interconnected ISP networks worldwide.

Packets: data travels in small, independent pieces

Rather than sending a file or webpage as one continuous stream, the sending device breaks it into small units called packets. Each packet is labeled with source and destination IP addresses and can travel its own path across the network, with routers reading each packet's destination and deciding the next hop along the way.

This approach, known as packet switching, was theorized by Leonard Kleinrock at MIT in a July 1961 paper, according to the Internet Society's history of the internet. In 1965, Lawrence G. Roberts built an early wide-area computer network connecting computers over a telephone line, demonstrating that packet switching outperformed the traditional circuit-switched telephone model for moving data.

TCP/IP: the common rules networks agree to follow

For packets from different, independently designed networks to interoperate, a shared protocol was needed. Robert Kahn, who joined the U.S. Defense Advanced Research Projects Agency (DARPA) in 1972, set out open-architecture networking principles and worked with Vint Cerf, then at Stanford University, starting in spring 1973 to design what became TCP/IP, the Internet Society states. Their design was circulated as a working paper in September 1973 and published in refined form in 1974.

Kahn's founding rules, as described by the Internet Society, were that each network could keep its own internal design, communication would happen on a best-effort basis without guarantees, connecting gateways would stay simple, and no single authority would exercise global operational control. That decentralized design is why no one company or government administers the internet as a whole.

DNS: translating names into addresses

Typing a domain name such as icann.org into a browser does not communicate directly with any machine — computers need the numerical IP address instead. The Domain Name System (DNS) performs that translation and is described by the Internet Corporation for Assigned Names and Numbers (ICANN) as "the address book of the Internet."

ICANN, the nonprofit organization that coordinates the global domain name system, explains that every internet-connected device has a unique IP address, and DNS lets people use readable domain names instead of memorizing number sequences. A domain name consists of two or more text segments separated by dots, and when a browser looks one up, DNS returns the matching IP address so the connection can proceed.

How the data physically crosses continents

Once past local networks, long-distance internet traffic between continents overwhelmingly travels through undersea fiber-optic cables rather than satellites. Industry data provider TeleGeography states that submarine cables account for more than 99% of intercontinental data traffic.

TeleGeography also notes a caveat worth attributing precisely: that widely repeated figure traces back to U.S. Federal Communications Commission (FCC) data from 2013 showing satellites made up just 0.37% of U.S. international capacity at the time, and the FCC no longer requires satellite operators to report updated circuit figures. TeleGeography says the general claim about submarine cables is directionally accurate but cannot currently be verified with precise, up-to-date global figures.

Putting it together

When someone opens a webpage, a device first queries DNS to resolve the site's domain name into an IP address. The device then sends packets addressed to that IP address; routers inside ISP networks forward each packet, potentially over undersea cables if the destination is on another continent, until the packets reach the destination server. The server sends its response back as packets, which the original device reassembles into the page the user sees.