In this first module, we'll take an in-depth look at the TCP/IP protocol, the cornerstone of modern digital communications. We will discuss its origins, its fundamental principles, and the addressing system it uses, which is essential for ensuring the flow of information between connected devices.

We'll also detail the main components that structure this model, and explain how they interact to form an operational, reliable and scalable network. But first, it's essential to go back to the concept of a network.

Etymologically, a network refers to a set of points linked to one another, forming an interconnected structure. In telecommunications and computing, this definition translates into a group of devices (computers, routers, switches, access points, etc.) capable of exchanging data via physical or wireless media. A network thus enables the continuous or intermittent flow of information, depending on requirements, on the protocols in use and on the nature of the deployed architecture.

Over time, several classic topologies have been developed to meet different needs for cost, performance, resilience, and ease of maintenance. These include:

In a ring topology, devices are connected in a closed loop: each station is linked to the next, and the last one connects back to the first. In this setup, every device acts as a relay, passing data along to the next link. Depending on the network type, information can circulate in one direction only, or in both.

The advantage of this arrangement lies in the simplicity of its cabling, and the absence of dependence on any central equipment. However, the continuity of the entire network depends on the health of each individual element: the failure of a single station can interrupt the entire communication system. This is why redundancy or bypass mechanisms are often put in place.

The tree network, or hierarchical topology, is modeled after the structure of a family tree. It consists of successive levels: a root node at the top connects to several lower-level nodes, which may themselves connect to other nodes, and so on.

This hierarchical layout works particularly well for large networks that need a clear division of responsibilities and segmented management. However, it also makes the network vulnerable to the failure of higher-level nodes: losing the root or a main branch can cut off entire sections of the infrastructure.

In a bus topology, all devices share the same transmission medium, typically a coaxial line or optical fiber. Each unit is passively connected, meaning it doesn't actively modify the signal, and it can send or receive data over this shared channel.

The bus topology's main advantage is low installation cost, thanks to simplified cabling. However, in older coaxial-based implementations (Ethernet 10BASE2/10BASE5), disconnecting or losing a single station could disrupt or even halt all traffic, as the bus electrical continuity and termination impedance would no longer be maintained. Having a single physical medium is also a critical weakness: any break or fault stops communication for the entire network.

The star topology, also known as "hub and spoke", is the most common today, especially in home and office Ethernet networks. Here, all devices connect to a single central device.

This layout makes management and maintenance easy: if one peripheral device fails, the rest of the network remains unaffected. The downside is that the central device is a single point of failure: if it goes down, communication stops everywhere. Cable quality and link lengths must also be carefully considered to maintain good performance.

Note: there are still networks organized in a linear, bus-like topology, where equipment is connected one after the other. This solution, although inexpensive to deploy, has the major drawback that a single break isolates some of the hosts, splitting the network into independent subsets.

The mesh network is designed for maximum redundancy: every device is directly connected to every other device. This ensures service continuity even if multiple links or devices fail, as traffic can be re-routed along alternative paths.

The trade-off is that the number of connections to be established increases rapidly with the number of terminals. For N connection points, N × (N-1) / 2 separate links are required, making this topology expensive and complex to deploy. It's therefore used mainly in critical networks requiring very high availability, such as certain parts of the Internet or sensitive industrial systems.

Other variations exist, such as grid or hypercube networks, which are designed for specialized needs in distributed computing or parallel processing.

On a global scale, the Internet is a massive interconnection of networks using diverse topologies, unified by common addressing (IPv4 and IPv6) and a collection of standardized protocols defined by the IETF (Internet Engineering Task Force). This diversity means the Internet follows no single topology: its structure is flexible, scalable and independent of the logical addressing scheme that makes it usable.