Let's be honest, humans aren't great at memorizing long strings of numbers, whether in binary or decimal form. This challenge becomes even greater with IP addresses, which can be complex and a single IP address can sometimes mask multiple addresses, especially when techniques like NAT or virtual hosting are involved.

To make things easier, the Application layer uses a system that links an IP address to a logical, human-readable name. This is the role of DNS (Domain Name System), a massive, hierarchical, distributed directory that matches readable domain names to IP addresses. The system is based on a set of protocols and services. The most widely used DNS server software is BIND (Berkeley Internet Name Domain), an open-source software package that references much of the Internet’s DNS infrastructure.

The core idea behind DNS is simple: for any connected service, whether a website, mail server, or another network service, there is a record mapping a domain name to one or more IP addresses. This works in two directions:

A domain name is always structured hierarchically, with each level separated by a dot: the full name is called FQDN (Fully Qualified Domain Name). The rightmost part is the TLD (Top Level Domain) such as .com, .org or .fr. The left-most part designates the host, i.e. the specific machine or service linked to the IP address.

The DNS system is designed as a tree of zones. A zone is a section of the domain namespace managed by a specific DNS server. A single zone can contain multiple subdomains, which may themselves be delegated to other zones managed by different servers. Administrators are responsible for maintaining their zones: handling updates, delegations, and overall management.

This structure allows not just pointing to a main domain (e.g. example.com), but also fine-tuning records for individual hosts (www, mail, ftp, etc.). In the early days of networking, this mapping was handled with static files like (/etc/hosts on Unix systems), but such a method quickly became impractical for a fast-growing, interconnected Internet.

It's important to understand that a DNS server may only serve a limited scope. For example, a company's internal DNS server might not be directly accessible from the Internet. If this DNS is not configured to forward queries, or does not have a trusted relationship with other servers, some queries will fail: neither the name nor the IP address can then be resolved outside the defined zone.

DNS also plays a role in email routing. For example, a MX (Mail Exchange) record specifies the mail servers responsible for receiving e-mails for a given domain. These records define priorities (weighting factor) and failover solutions. The zone file of a DNS server must contain a SOA (Start Of Authority) record, which designates the server as the official source of information for that zone.

Thanks to its hierarchical, distributed structure, DNS remains a cornerstone of the Internet, allowing users to access services through clear, memorable domain names instead of long, technical IP addresses.

In the next chapter, we'll explore another fundamental concept: Ethernet addresses, also known as MAC addresses, which ensure data delivery at the physical layer of local networks.