Network programming in Linux revolves around the concept of sockets, which serve as the abstract interface between application processes and the underlying network protocols. Understanding these foundational elements is crucial before diving into specific protocol implementations.

1.1 The Nature of Sockets and Inter-Process Communication

A socket is an abstract entity representing one endpoint of a communication link. It's a software construct that functions similarly to a file descriptor, allowing applications to perform input/output operations across a network. When an application creates a socket, it essentially requests an endpoint for communication from the operating system kernel. This endpoint is then used to send and receive data. Sockets enable Inter-Process Communication (IPC), specifically over a network, allowing processes on different machines (or even the same machine) to exchange data.

Sockets are characterized by their:
- Domain (or Address Family): Specifies the network protocol family used.
- AF_INET: For IPv4 Internet protocols.
- AF_INET6: For IPv6 Internet protocols.
- AF_UNIX (or AF_LOCAL): For local communication between processes on the same machine.
- Type: Defines the communication semantics.
- SOCK_STREAM: Provides a connection-oriented, reliable, sequenced, and unduplicated flow of data (e.g., TCP). Data is treated as a continuous stream of bytes.
- SOCK_DGRAM: Provides a connectionless, unreliable datagram service (e.g., UDP). Data is sent in discrete, independent packets.
- SOCK_RAW: Allows direct access to the underlying network protocols, bypassing higher-level protocols like TCP/UDP. (More advanced, typically for network analysis tools).
- Protocol: Specifies a particular protocol within the chosen domain and type (usually set to 0, letting the system choose the default protocol).

1.2 The Client-Server Model: Roles and Communication Principles

The vast majority of network applications adhere to the client-server model.
- Server: A server process is typically passive. It creates a socket, binds it to a specific well-known IP address and port number (making itself discoverable), and then listens for incoming connection requests (for TCP) or incoming datagrams (for UDP). When a client initiates communication, the server accepts the connection (TCP) or receives the datagram (UDP) and then provides the requested service or data. A server often needs to handle multiple clients concurrently.
- Client: A client process is active. It creates a socket and then actively attempts to establish a connection with a specific server (for TCP) or sends a datagram to a specific server address (for UDP). Once communication is established or data is sent, the client consumes the service or data provided by the server.

Key Concepts in Network Addressing:
- IP Address: A numerical label (e.g., 192.168.1.1 for IPv4, 2001:0db8::1 for IPv6) that uniquely identifies a network interface of a host on an IP network.
- Port Number: A 16-bit number (0-65535) that identifies a specific application or service running on a host. Well-known ports (0-1023) are reserved for standard services (e.g., HTTP: 80, HTTPS: 443, FTP: 21, SSH: 22). Registered ports (1024-49151) are assigned by IANA for specific services. Dynamic/Private ports (49152-65535) are often used by clients or temporary services.
- Socket Address Structure (sockaddr_in / sockaddr_in6): Sockets require a way to represent network addresses. The struct sockaddr_in (for IPv4) and struct sockaddr_in6 (for IPv6) are used to store IP addresses and port numbers. These structures must be cast to the generic struct sockaddr* when passed to socket system calls.