Networking Software - Introduction

In this section I tried to give answers to the following questions:

How does a typical network software achitecture look like?
To run a network software is needed. This software for instance performs the addressing of packages, error detection, avoidance of data collisions and -losses, the routing of data, encryption, and much, much more.
Because of the enormous variety of services a network software has to offer, the work to be done is organized as a series of layers, each one built upon the one below it. Every layer offers certain services to the layer above but it hides the actual implementation of these services.
Using a layered software architecture is the only way to keep network utilities somewhat simple: Because it is agreed upon what services the layers offer, networking software is standardized referring to input- and output formats, addressing, speed, and so on.
Table: a 5-layered network architecture
Layer 5 Peer-to-peer protocol (5th layer)
 <----------------------------------------> 		Layer 5
Layer 4 Peer-to-peer protocol (4th layer)
 <----------------------------------------> 		Layer 4
Layer 3 Peer-to-peer protocol (3rd layer)
 <----------------------------------------> 		Layer 3
Layer 2 Peer-to-peer protocol (2nd layer)
 <----------------------------------------> 		Layer 2
Layer 1 Peer-to-peer protocol (1st layer)
 <----------------------------------------> 		Layer 1
Physical medium
Example:
Email encryption is a typical service of network software. But it is only about what data will be sent, not how. So the encryption routine passes the result of encoding to the layer below for transmission.
At the receiver's place the message is passed up the layers until it reaches the decryption program. This one makes the message readable for the recipient and passes it up to the program that displays it.

What are reference models and which of them are currently in use?
A reference model is a layered network architecture to which (ideally) all companies refer. It has a specified amount of layers, the protocols are agreed upon, and each layer got certain services attached. Graphically, it is a structure like the one above, but on a less abstract level.
There is not at all an agreement upon what reference model to use internationally; the amount of layers to use and their names are under discussion, as well as is the categorization (which services are part of what layer). During the time, however, three models grew to extraordinary importance:
© 2000 Peter Gallert, last updated on 24 January 2001