11.1 Data transmission

Using the analogy of shipping containers, Session 1 introduced you to the idea of a layered approach to data transmission where each layer is responsible for a different aspect of the transport and delivery of the data. The important concept to take away from this is that the message content is completely independent of the mode of transport used and the services associated with it.

In Session 1 you were also introduced to the four layers of the TCP/IP protocol stack, which is the suite of protocols used on the internet and most other computer networks. In Session 3 you learned that the data created by the users at the Application layer is sent down the stack to the next layer, the Transport layer, where it is segmented into chunks of the same size called data units before being passed down the stack to the next layer. Each layer adds some information that enables it to do its job. This is called encapsulation. The Internet layer is responsible for getting data units across networks – sometimes via many different networks. The information that’s added here includes the IP addresses and subnet masks of the source and destination networks. Routers work at this level and use IP addresses and subnet masks to get information from network to network.

But why do we need to go to all this trouble of encapsulation? Why don’t we just launch data units out onto ‘the internet’, with the IP address uppermost, and have them successfully delivered? To answer this, we need to think about what the internet is. You might have heard it described as a ‘network of networks’ which makes it easier to understand.

The Internet is made up by interlinking many different sorts of networks – so many that it is impossible to draw a full map of them. The network technologies work together to cross geographical, system and ownership boundaries.

For example, the telecomms companies had nationwide and international high speed packet-based networks back in the sixties and seventies. These used addressing systems and technologies that were particular to that industry. Offices and businesses in the early 80s had small networks to join computers to each other, to printers and to storage devices using protocols like AppleTalk, Ethernet and token ring. Often companies would link their own remote offices and sites with networks connected together through leased lines that used different technologies. All these networks were like islands where different languages were spoken. There was no obvious way to get data from one such network to another, because initially people didn’t see the need for data to go outside its local network.

What IP addressing and routers give us are a way of getting data from one such ‘island’ to another. But on the island itself, data has to be carried by the local form of transport. That’s why IP data has to be encapsulated, and the outer layer of the encapsulation has to use whatever addressing system is particular to the network the data is being carried across. This is why we speak of ‘lower-level’ and ‘higher-level’ networks and addressing. The lower-level addressing is meaningful only on a local network, whereas the higher-level addressing has to be interpretable on external networks. By the way, a ‘local network’ isn’t necessarily a small one.

So, you’ve met the TCP/IP protocol suite, but there is another layered model that is commonly referred to in data networks. This is known as the Open Systems Interconnection model, abbreviated to OSI. In this model, seven layers are defined and, though the model is only conceptual, you will often encounter network functions being identified with OSI layers instead of TCP/IP layers. The OSI model is much more detailed. For example, the Physical layer (layer 1) deals with whether the physical medium is cable, optical fibre or wireless. The Data Link layer (layer 2) deals with the switching of data in the local network and the Network layer (layer 3) deals with routing data across the internet.

Fortunately there is quite a straightforward mapping between TCP/IP and OSI as you can see here. So when you come across functions identified as occurring at layer 3 of the OSI model, this is actually referring to the Network layer (or the Internet layer in the TCP/IP model).

Switches operate at the Network Access layer of the TCP/IP stack and the Data Link layer (layer 2) of the OSI model. Their purpose is to connect devices within a local network. Routers operate at the Internet layer of the TCP/IP stack and the Network layer (layer 3) of the OSI model. Their function is to connect networks. You can think of a router as being a gateway from one network to another. Within the network itself the router doesn’t really have any function related to delivery. That’s handled by the switch, and it’s why you can ping from one computer to another on the same network even if the computers aren’t configured to know about the router. Routers, therefore, continually direct incoming packets to other routers until a packet arrives that is destined for the local network that the router is a gateway to. Also, packets on the local network come to the router when their destination is outside the originating local network.