Progress pill
TCP/IP protocols

Service primitives

IP networks - From Theory to Practice

Service primitives

  • Layered architecture and exchange organization
  • Service primitives and data units
  • Part summary

Layered architecture and exchange organization

As we've seen, services are the concrete implementation of the protocols we've described so far. While the TCP/IP model differs from the OSI model, it adopts the same layered approach: each layer is designed to perform a specific function and to provide services to the layer directly above it, resulting in a modular, robust, and easily maintainable architecture.
Each layer builds on the capabilities of the one below it, and in turn provides the layer above with a consistent Interface for managing data. In this architecture, every layer has its own data structures, carefully defined to ensure perfect compatibility with the other layers. This compatibility is essential for smooth, reliable, and clear communication from one endpoint to another.
Two key aspects govern these exchanges:
  • Vertical aspect: the relationship between one layer and the one above or below it (from layer N to layer N+1, and vice versa).
  • Horizontal aspect: the interaction between remote applications, i.e., the dialogue between a client and a server, in either direction.
The layered architecture follows the principle that each layer processes only the information within its scope: data structures, headers and control mechanisms vary from one layer to another, but together they form a coherent system, ensuring data is gradually routed to its final destination.
Reminder: Specific terminology is used to describe the data units exchanged between layers:
  • message for the Application layer,
  • segment for the Transport layer (TCP),
  • datagram for the Internet layer (IP),
  • frame for the Network Access layer.
The table below summarizes the terms for TCP and UDP contexts:
TCP/IP LayerUnit Name (TCP)Unit Name (UDP)
Application LayerStreamMessage
Transport LayerSegmentPacket
Internet LayerDatagramDatagram
Network Access LayerFrameFrame

Service primitives and data units

At the core of this system are service primitives, which act as communication interfaces. These primitives function like service desks, listening on reserved specific ports and allowing processes to establish, maintain, and terminate network connections in a controlled way. While protocols organize the format and transmission of data across the network, it is the services and their primitives that provide the vertical link between the layers.
By combining the horizontal aspect (communication between distributed applications) with the vertical aspect (internal interactions between layers), the TCP/IP model delivers a complete, scalable architecture. Overlapping these two perspectives gives a clear overview of how data is exchanged in structured network communication.

Part summary

In this first major section, we explored the core architecture that governs the configuration and operation of today's Internet-connected networks. This architecture is based on a four-layer model, inspired by the OSI model, and built around the TCP/IP protocol suite, the backbone of modern communications. We saw that TCP, with its connection-oriented approach, ensures reliable transfers, while UDP, lighter and faster, is preferred when speed is more important than reliability.
The proper functioning of this model relies on the implementation of protocols through service primitives. These ensure the link between layers, enabling data processing to be adapted to the specific requirements of each level, from transport to application, including Internet and network access. This modular approach makes the system both flexible and robust.
IP addressing is another cornerstone of this infrastructure. Every connected device is identified by a unique IP address, taken from an address space organized into classes (from A to E). Some of these addresses are reserved for special purposes, such as local loopback or multicast, while others, known as "private addresses", are not routed over the Internet without translation (NAT). This classification enables a logical, hierarchical organization of networks.
We also examined the concept of subnetworks, which makes it possible to divide a network segments to better manage IP resources and optimize data flow. While manual subdivision using subnet masks remains an important principle, it has largely been modernized thanks to CIDR (Classless Inter-Domain Routing). This method has transformed address management by enabling more flexible and rational allocation of IP ranges, while reducing the size of routing tables.
By mastering these concepts - layers, protocols, service primitives, addressing and subnetting - you gain a solid foundation for understanding the technical workings of modern networks, and for efficiently configuring a network infrastructure to meet today's needs.
In the next section, we'll take a closer look at IPv4 addressing.
Quiz
Quiz1/5
Which data unit is used in the Internet layer of the TCP/IP model?