A wire can get data from one place to another. However, you already know that the Internet can get data to many different places, distributed all over the world. How does that happen?
The different pieces of the Internet are connected by a set of computers called routers, which connect networks together. These networks are sometimes Ethernets, sometimes token rings, and sometimes telephone lines, as shown in Figure 3-1.
The telephone lines and Ethemets are equivalent to the trucks and planes of the Postal Service. They are means by which mail is moved from place to place. The routers are postal substations; they make decisions about how to route data ("packets"), just like a postal substation decides how to "route" envelopes containing mail. Each substation or router does not have a connection to every other one. If you put an envelope in the mail in Dixville Notch, New Hampshire, addressed to Boonville, California, the Post Office doesn't reserve a plane from New Hampshire to California to carry it. The local Post Office sends it to a substation; the substation sends it to another substation; and so on, until it reaches the destination. That is, each sub-station only needs to know what connections are available, and what is the best "next hop" to get a packet closer to its destination. Similarly, with the Internet: a router looks at where your data is going and decides where to send it next. It just decides which pipe is best and uses it.
How does the Net know where your data is going? If you want to send a letter, you can't just drop the typed letter into the mailbox and expect delivery. You need to put the paper in an envelope, write an address on it, and stick a stamp on it. just as the Post Office has rules about how to use its network, the Internet has rules about how to use it. The rules are called protocols. The Internet Protocol (IP) takes care of addressing, or making sure that the routers know what to do with your data when it arrives. Sticking with our Post Office analogy, the Internet Protocol works just like an envelope (Figure 3-2).
Figure3-2. IP envelopes
Some addressing information goes at the beginning of your message; this information gives the network enough information to deliver the packet of data.
Internet addresses consist of four numbers each less than 256. When written out, the numbers are separated by periods like this:
192.112.36.5
128.174.5.6
(Don't worry; you don't need to remember numbers like these to use the network.) The address is actually made up of multiple parts. Since the Internet is a network of networks, the beginning of the address tells the Internet routers what network you are part of. The right end of the address tells that network which computer or bost should receive the packet.* Every computer on the Internet has a unique address under this scheme. Again, the Postal Service provides a good analogy. Consider the address "50 Kelly Rd., Hamden, CT." The "Hamden, CT" portion is like a network address; it gets the envelope to the right local Post Office, the Post Office that knows about streets in a certain area. "50 Kelly Rd." is like the host address; it identifies a particular mailbox within the Post Office's service area. The Postal Service has done its job when it has delivered the mail to the right local office, and when that local office has put it into the right mailbox. Similarly, the Internet has done its job when its routers have gotten data to the right network, and when that local network has given the data to the right computer, or host, on the network.
For a lot of practical reasons (notably hardware limitations), information sent across IP networks is broken up into bite-sized pieces, called packets. The information within a packet is usually between 1 and about 1500 characters long. This prevents any one user of the network from monopolizing the network and allows everyone to get a fair shot. it also means that if the network isn't fast enough, as more people try to use it, it gets slower for everyone.
One of the amazing things about the Internet is that, on a basic level, IP is all you need to participate. It wouldn't be very friendly but, if you were clever enough, you could get some work done. As long as your data is put in an IP envelope, the network has all the information it needs to get your packet from your computer to its destination. Now, however, we need to deal with several problems:
Most information transfers are longer than 1500 characters. You would be disappointed, indeed, if the Post Office would only carry postcards, but refused anything larger.
Things can go wrong. The Post Office occasionally loses a letter; networks sometimes lose packets, or damage them in transit. Unlike the Post Office, we'll see that the Internet can deal with these problems successfully.
(*Where the network portion ends and the host portion begins is a bit complicated. It varies from address to address based on an agreement between adjacent routers. Fortunately, as a user you'll never need to worry about this; it only makes a difference when you're setting up a network.)
Packets may arrive out of sequence. If you mail two letters to the same place on successive days, there's no guarantee that they will take the same route or arrive in order. The same is true of the Internet.
So, the next layer of the network will give us a way to transfer bigger chunks of information, and will take care of the many "distortions" that can creep in because of the network.