Computer Network Architecture

Network communication looks like “moving data from A to B”, but the hard part is how to implement reliable, ordered, multi-party communication over shared bandwidth on unreliable physical media. Layering exists to solve exactly that problem: every layer offers services to the layer above and does not care how the layer below realizes them, so a large amount of complexity is cut into modules that can evolve independently.

This chapter first establishes the basic concepts and the classification, so you know which parts a network is made of and which problems it solves; then it introduces performance metrics to make “fast” and “good” quantifiable. Next comes the layered structure, where the precise meaning of protocol, interface, and service matters — they draw the boundary of responsibility between layers. After that, OSI’s seven layers and TCP/IP’s four layers are compared, showing that the two are different cuts of the same idea rather than two unrelated systems. Finally the abstraction returns to reality: routers, switches, and hubs each operate at a specific layer of the model, and their roles follow from that.

After reading this chapter you should be able to explain why layering exists, how the OSI and TCP/IP models relate to each other, and where a given device sits in the model.