How the Internet works
(Jellybeans in the Park)

Imagine two kids sitting in a park with a length of tube and a few packs of jellybeans. The different coloured jellybeans represent different messages. A yellow jellybean means "here's a nice picture of the planet Mars". A white jellybean means "please send me something". One kid picks up the tube and sends a white jellybean down it to his friend. His friend gets the white jellybean, picks up the tube and sends a yellow jellybean back. Hey Presto, a picture of Mars.

That's the basis of the Internet. In reality the jellybeans are data packets and the tube is a telephone line. Your computer sends a white jellybean to, say, www.nasa.gov, and nasa.gov sends a sequence of jellybeans back that create a picture of Mars on your computer. You send a request, you get a Web page back in response.

Now lets have five kids in the park, with lots more tubes and jellybeans. Mum is very organised and sits David in the middle with the others evenly arranged around him. There are now four tubes, one from David to each of the other four kids. Charlie want something from Elizabeth, so he picks up a white jellybean and with a fine pen writes on it "To E from C" (he's very advanced for his age). He picks up his tube and sends this bean down to David in the middle. David picks it up, sees it's addressed to E, gets hold of his tube that leads to Elizabeth and forwards the jellybean to her.

Elizabeth sees the white bean is from Charlie. She picks up a whole bunch of beans and writes on them "To C from E, first bean", "To C from E, second bean", and so on. She then sends these down her tube to David, sitting in the middle. He reads they're for Charlie and sends them on. Charlie receives a little pile of jellybeans. He picks them up and puts them in the right order and he's got his message from Elizabeth. It reads "znlpff" (Elizabeth hasn't quite got the hang of this yet).

Routers

David has become, in Internet terms, a router. All he does is look at the addresses on the beans and send them on. That's exactly how the Internet works, a lot of routers in the middle transferring jellybeans from one place to another.

Lots of other mums think this is a great way to keep their kids entertained, so they join in. Soon there are fifty kids in an uneven circle spread around the park, and in the middle of them are another thirty kids doing David's job, acting as routers. The kids closest to the centre don't have any direct connection with anybody on the edge of the circle, all their tubes simply go to another middle kid who's also acting as a router. And the kids fairly near the edge of the circle have a mix of connections, some to kids on the edge and some to kids in the middle.

Charlie wants to hear from Elizabeth again and see if she can send a sensible message this time, but she's way over the other side of the big circle, so he knows his white bean is going to have to go through a lot of routers. He also knows that most of the kids in the middle don't know who Elizabeth is - they've never met her. So this time he writes on his bean, "To E from C - and by the way, E is sitting in the south-east part of the circle."

He sends this bean down his tube to a kid called Fiona sitting just inside the edge of the circle, who picks it up and reads it. Fiona has no idea who Elizabeth is, but knows that south-east is vaguely in that direction over there, so she sends the bean to the kid sitting roughly south-east of her.

This kid forwards it in turn to whoever is sitting south-east of him, until it arrives at a kid who is sitting very close to Elizabeth. An important bit of the system comes into play here, which is that all the router-kids near the edge definitely know the names of the closest kids on the circumference of the circle.

So, the kid close to Elizabeth recognises that the bean is for her, and passes it on. And this time she writes on her first reply bean, "To C from E, first bean - and by the way, C is sitting at the north-west of the circle."

Of course Charlie and Elizabeth aren't the only kids communicating on this primitive form of the Internet. All the other kids on the edge are doing similar things. So there are hundreds of jellybeans passing through the middle in all kinds of directions. It's chaotic, but in an organised kind of way, so that everybody does eventually get the beans that are addressed to them, even if it takes a while.

Resilience

Now we go one step further. Some of the kids have got plenty of white beans but want to collect pretty coloured ones. They call themselves home computers. Other kids have big stacks of multi-coloured beans that they generously want to give away. These kids call themselves Web servers. Most of the traffic through the middle is made up of coloured beans sent from Web servers to home computers in response to white bean requests.

This is pretty much how the Internet works. Lots of home computers wanting to receive information, and lots of Web servers wanting to supply it. And in the middle, a host of routers owned by telecom companies and ISPs and universities and government agencies and other people with technical knowledge. The jellybeans themselves are packets of digital data.

Of course, amongst all these kids there has to be one bad apple. There's a fat kid near the middle who reckons this is a stupid game. When he thinks nobody's looking, he eats the beans instead of sending them through. He's gobbled down hundreds and won't be able to finish his dinner tonight.

Amazingly, the system copes with this. Remember, the beans have their order written on them - first bean, second bean, third bean and so on. So when David puts the beans he's received from Elizabeth in order, he notices that bean seven is missing. He waits a while, still no bean seven. Finally he gets out a special pink bean and writes on it "The fat sod in the middle has eaten bean seven, can you send me another one?"

Elizabeth obliges. The whole transaction is slower than it should have been, but is eventually completed.

But barely is that little crisis over when something even worse happens. One of the kids in the middle gets stung by a wasp. He starts bawling his head off. Mum arrives and takes him in her arms and carries him off to comfort and safety.

Disaster! We're a router down. What will happen to all the beans that used to pass through it?

Simple. The kids close to the empty patch of grass where the wasp sting occurred see that it's pointless using the tubes that lead there, because the router's gone. So instead they send their beans in a slightly different direction, maybe west when the bean should really be going south-west, and so on. The kid to the west then sends the bean off in the right direction again, south-west, and it's effectively skirted the empty patch of grass. This slows the mechanisms down a bit, but the beans still get where they're going.

The real Internet was designed to be resilient to something slightly worse than a wasp sting, it was developed by the military to cope with nuclear war. The idea was that even if hundreds of routers were wiped out by scores of bombs, the Internet would still work and important messages would get through.

DNS

We can go a couple of steps further with the jellybean analogy. Let's imagine that all the kids sitting on the edge of the circle are French, and all the kids in the middle are Chinese. All the messages on the beans are written in French and the Chinese kids can't understand them. They don't need to understand everything that's written on the bean, but for the system to work, they do need to understand the direction and some of them need to understand the receiver's name.

So a new set of kids join in as translators. They sit near the edge of the circle and introduce a system of orange beans. Whenever anybody wants to send a white bean, they have to send an orange bean first. The routers know that orange beans must go to a translator, and the translators turn the French that's written on these into Chinese and send it back. The outer circle kids now have Chinese script they can write on their white beans (they're very gifted kids). All the routers can read this and the white bean reaches its destination.

In the real world the translators are called Domain Name Servers (DNS). On the real Internet they convert the letters of domain names into numbers, since computers easily understand numbers but the alphabet is a bit of a mystery to them. The numbers are called IP addresses (Internet Protocol addresses). The IP address for www.nasa.gov is 198.116.142.34.

Everybody connected to the Internet has an IP address. It's very similar to a street address and no two are the same. The IP address effectively incorporates the direction for a jellybean as well as the recipient's name. For example, the first three numbers in an IP address correspond roughly (but not exactly) to the recipient's country.

When you type www.nasa.gov into your browser, your first bean is sent to a Domain Name Server. This translates www.nasa.gov into 198.116.142.34 and sends that information back to you. Now you can address your white bean in a way the routers can fully understand, and send it off. When the nasa.gov server sends its coloured beans back, they don't have to go through the DNS system, because the return address you sent with your white bean was already in numerical format.

Actually, your return address isn't usually determined by your computer (though it may be if you're using a networked computer at work), it's allocated by your ISP (Internet Service Provider). When your computer rings up your ISP to get an Internet connection, you borrow one of your ISP's IP addresses for your Internet session. The ISP then receives all your jellybeans and forwards them on to you.

Packets

Perhaps the last useful bit of the jellybean analogy is the size and nature of the beans themselves. Each jellybean corresponds to an Internet data packet, and these packets are self-contained and have address information (and their order) included in their headers, or labels. Inside the packet is some digital data that your computer can understand. When it strings together the data from a complete sequence of packets, it finishes up with an HTML page or a graphic or possibly an email. This then shows up on your screen.

Although jellybeans are fairly consistent in size, Internet data packets are not. They can even change size as they travel through the routers of the Internet, which may split them into smaller packets or combine them into larger ones.

The single white jellybean used in our analogy might start off in reality as twenty packets sent from your computer, and could have turned into ten or fifty by the time it reaches the Web server at the other end.

In terms of file size, a sequence of packets is generally bigger than the data it carries. So a graphic on an HTML page that's 5k in size might be carried by a sequence of packets totalling 7k altogether. This is partly because the data may be transformed into a different format before it's put inside the packets, but mainly because the packets carry header information, including IP addresses and the order of reassembly, and this content is an overhead. That's why if you measure the data flowing into your computer through the Internet and then measure the amount that's showing up on your screen, the two don't correspond. There's invariably a lot more data coming in than you get to see, and that's mainly down to data packet overheads.

And that's enough technical mumbo-jumbo. Now I'm going down to the local park with a giant box of jellybeans and a bag full of plastic tubes. I've made arrangements for bail.

What every Web site can find out about you

TinHat.com home