2.3 Networks of networks

Near the start of this session, I showed you that the gateway has a side that faces the internet. You can see that the IP address of the internet-facing side of the gateway is very different from the one on the side that faces the local network. The internet-facing side doesn’t have the usual three numbers, 192.168.0, at the start, but instead has an address starting with 99. This means that the internet-facing side of the gateway is on a different network from the side that faces the local network.

One of the jobs of a router is to transfer data from one network to another, and a gateway is, among other things, a router. Notice also that the subnet mask on the internet-facing side of the gateway is different from the ones we’ve already seen. This one is 255.0.0.0, which is very different from 255.255.255.0.

Now the gateway address shown here, 99.0.0.1, doesn’t refer to this gateway, the one that we’ve been looking at on the home network It refers to the gateway at the internet service provider. The internet service provider is the company that provides a domestic customer, or a business customer, with access to the internet.

The internet service provider has a router, or possibly several, serving customers and their own home networks. So, the gateway on our home network is also part of the internet service provider’s network. In our example, the internet-facing side of the home gateway has the address 99.0.0.2, and we could guess that other home networks served by this internet service provider would have addresses such as 99.0.0.3, 99.0.0.4, and so on.

But the internet service provider’s router is also part of another network, usually a network of core routers, which might serve other internet service providers, or other large-scale users. Core routers are high-speed routers carrying traffic on the internet backbone.

So you can see the internet is a network of networks. Because it’s a network of networks, there’s often more than one route from A to B. This gives the internet some of its robustness, because if part of the internet stops working properly, there’s usually an alternative route that can be used.

The internet addressing system is really a system for getting data from one network to another. It doesn’t specify a route, just a destination, and there’s no guarantee that all the data packets of a data transfer will go by the same route.

We’ve already seen that a network address consists of a part that represents the network and a part that represents the device. As a packet of data travels from network to network, only the network part of the address is used, because routers transfer data from network to network not from device to device. That’s why there always has to be a subnet mask, that travels with the data packet, so that routers along the way know which part of the destination address is the network part. Only when the packet gets to its final network does the device part of the address get consulted, and on the final network it’s actually a switch that makes the delivery, although on home gateways the switch is built in to the gateway.

This picture is complicated a bit by the fact that some IP addresses can only be used on private networks. The 192.168 class of addresses belongs to those that can only be used on private networks. We’ll see more about private networks later.

But there has to be a certain amount of trickery to get data with private addresses across the internet. And we’ll look at that later.

When you think of a packet of data travelling from network to network, you have to think of it as travelling from router to router in a series of hops, as they are called. Usually a packets has a maximum number of hops specified, so if the maximum is reached before the packet gets to its destination, the packet is discarded. This prevents the internet getting clogged up with data that has never been delivered, for example if the destination network no longer exists. But it also means that delivery of a packet is not guaranteed.

Each router in the internet has to do a certain amount of processing work to transfer a packet from one network to another. The router has to look at the destination address, and it has and decide which router, of the ones it’s connected to, to forward the packet to. Sometimes a queue of data packets can build up at a router. So routers can introduce delay, and therefore contribute to the latency of the data transfer. Roughly speaking, the latency is a measure of the delay between sending some data and its arrival at the destination. On the internet, data can travel quite slowly compared with, say, radio, where transmissions travel at the speed of light.

The multitude of networks that make up the internet use many underlying communication technologies. Some use optical fibres, others use copper wires or coaxial cables. Sand some use wireless. In many cases the networks that make up the internet already existed before the internet came into widespread use. For example, in many countries there was already an extensive network of optical fibres carrying data traffic and telephone calls between major cities, before the internet arrived. So the introduction of the internet was not so much the introduction of a new set of networks, but the introduction of a new addressing system that could be used across existing networks.