Internet protocol suite

The Internet protocol suite, commonly known as TCP/IP, is a framework for organizing the communication protocols used in the Internet and similar computer networks. It helps computers talk to each other and share information. The main parts of this suite are the Transmission Control Protocol (TCP), the User Datagram Protocol (UDP), and the Internet Protocol (IP). These protocols work together to make sure data gets from one place to another correctly.

Originally, this networking model was called the Department of Defense (DoD) Internet Architecture Model because the research was funded by the Defense Advanced Research Projects Agency of the United States Department of Defense. It provides ways for data to be sent, received, and moved across networks. This is done by breaking data into small pieces called packets and sending them from one computer to another.

The Internet protocol suite organizes its functions into four main levels or abstraction layers. These layers help sort out all the different tasks needed for networking. Starting from the closest to the hardware, these layers are the link layer, which handles communication within a single network; the internet layer, which connects different networks together; the transport layer, which manages communication between specific computers; and the application layer, which deals with programs and applications that use the network.

Today, the rules and standards for the Internet protocol suite are kept up to date by the Internet Engineering Task Force (IETF). This suite was developed before another well-known model called the OSI model, which is a broader framework for all kinds of networking systems. Because of its importance, TCP/IP is the foundation of how most of the Internet works.

History

Further information: History of the Internet

The Internet protocol suite began with research in the late 1960s, funded by the Defense Advanced Research Projects Agency (DARPA). This research led to the creation of the ARPANET in 1969. Researchers like Steve Crocker started developing ways for computers to talk to each other.

In the 1970s, Bob Kahn and Vinton Cerf worked on new ways to connect different networks. They created a system where networks could hide their differences and communicate through a common set of rules. This led to the development of the Transmission Control Protocol and later split into the Internet Protocol (IP) and Transmission Control Protocol (TCP).

By 1983, Internet Protocol version 4 (IPv4) was used widely, and the system became known as TCP/IP. Over time, TCP/IP was adopted by many companies and became the standard way computers connect to each other online.

Key architectural principles

The Internet protocol suite follows several key principles to help it work well. One is the end-to-end principle, which originally meant that most of the smart parts of communication happened at the beginning and end, with the middle parts just moving data quickly. Over time, things like firewalls and caches have changed this idea a bit.

Another important principle is robustness: senders should carefully follow the rules, but receivers should be ready to accept messages even if they aren’t perfect. This helps keep things working even when there are small mistakes.

The suite also uses encapsulation, which means wrapping data in layers, like putting a letter in an envelope. This helps different parts of the network understand and move the data properly. The main layers are:

• The application layer is where programs like web browsers talk to each other.
• The transport layer makes sure data gets from one computer to another, using protocols like UDP and TCP.
• The internet layer handles moving data between different networks, mainly using IP.
• The link layer deals with how computers talk to each other on the same local network.

Main article: Communication protocol § Software layering

End-to-end principle

Robustness principle

Encapsulation

Application layer

Processes

Reliable or unreliable

Client–server model

Peer-to-peer

Services

Transport layer

User Datagram Protocol

Connectionless

Transmission Control Protocol

Internet layer

IP addresses

Link layer

RFC

Link layer

The link layer helps computers talk to each other on the same local network, like in your home or school. It works with the hardware that connects devices, such as cables and Wi-Fi. This layer moves small pieces of information, called packets, between computers without needing a router.

Special parts of a computer, like the network card and its software, help prepare these packets for sending. They also turn Internet Protocol addresses into the special codes each device uses, called media access control addresses. The link layer is similar to Layer 2 in another networking model called the OSI model.

Internet layer

The internet layer helps send data from one network to another. It uses a system called IP addressing to identify where data should go. This layer makes sure data can travel across many different networks, which is what lets the Internet work.

The main part of this layer is the Internet Protocol (IP). It has two versions: Internet Protocol version 4 (IPv4), which can handle about four billion addresses, and Internet Protocol version 6 (IPv6), which can handle many more. IPv6 was introduced in 1998 to solve the problem of running out of addresses.

Transport layer

The transport layer helps computers talk to each other by setting up ways for data to move between applications. It can work in two main ways: one where the connection is carefully set up first (like TCP), and another where data just sends out quickly without checking first (like UDP).

TCP makes sure data arrives correctly, in order, and without mistakes. UDP is faster but does not always check if every piece of data makes it through. Some apps, like watching live video or online games, use UDP because they need things to happen right away.

Application layer

The application layer includes the protocols that applications use to provide services or exchange data over network connections. Examples include Hypertext Transfer Protocol for web browsing, File Transfer Protocol for moving files, and Simple Mail Transfer Protocol for email. These protocols package data into units handled by lower layers, such as Transmission Control Protocol or User Datagram Protocol, which manage the actual data transfer across the network.

In the TCP/IP model, the application layer combines functions that other models split into separate layers. It works with the transport layer, which handles connections, but doesn’t worry about how data is formatted or presented. Common services use specific port numbers, like port 80 for HTTP, and these are managed by the Internet Assigned Numbers Authority. The application layer includes both user protocols, used for actual applications, and support protocols, which help the network infrastructure.

Layering evolution and representations in the literature

The Internet protocol suite grew and changed through many years of research and development. During this time, different groups worked on organizing the rules for how computers talk to each other, sometimes leading to different ways of grouping these rules. Efforts to bring these different ideas together helped create teaching tools for students and experts alike.

The table below shows some of these different models for organizing computer network rules. These models have anywhere from three to seven layers, and they come from textbooks and other sources.

Comparison of TCP/IP and OSI layering

See also: OSI model § Comparison with TCP/IP model

The TCP/IP model groups things a bit differently than the OSI model. While the OSI model splits things into separate layers like application, presentation, and session, TCP/IP just has one application layer on top of the transport layer. This means TCP/IP applications can use different ways to get their jobs done without needing to stick to a strict structure.

Some examples show how this works. For instance, the NFS application uses something called XDR for presenting data, which then uses RPC for making remote calls. RPC makes sure data gets sent safely, even though it uses UDP, which doesn’t guarantee safety on its own. Different people might describe where the TCP/IP model ends and the hardware begins, and some try to fit the OSI model's ideas into TCP/IP. But the group that builds Internet rules, the IETF, doesn’t worry too much about strict layers. They’ve even said that layering can sometimes cause problems.

Implementations

The Internet protocol suite works on many different types of computers and devices. It just needs to be able to send and receive small pieces of data called packets. Because of this, it has been used on almost every kind of computer system.

A basic version of this system includes several important parts: Internet Protocol (IP) for sending data, Address Resolution Protocol (ARP) for finding devices, Internet Control Message Protocol (ICMP) for messages between computers, Transmission Control Protocol (TCP) for reliable data transfer, User Datagram Protocol (UDP) for faster but less reliable data, and Internet Group Management Protocol (IGMP) for group communications. For newer versions of IP, it also includes Neighbor Discovery Protocol (NDP), ICMPv6, Multicast Listener Discovery (MLD), and often adds IPSec for extra security.