2.3.2 How TCP/IP works?
Messages sent over TCP/IP are called packets. Each packet of information sent over the Internet can be thought of as a letter. TCP/IP puts each letter in an envelope, addresses the envelope with “To” and “From” information , and sends the letter on its way. These packets are designed to be small, usually 1500 bytes or so.
Most things you send and receive on the Internet (e-mail messages and files) are not longer than the maximum packet size. Hence TCP/IP breaks the message up into packet/sized chunks, addresses each packet and sends them on their happy way.
Once at their destinations (actually getting them there is another story), TCP/IP reassembles the packets into one full continuous message. At the destination the IP verifies the labels, and pass them to the TCP. The TCP at the destination checks if all the packets have been receive. If any of the packets are missing, then it informs this to the source TCP and requests the packets to be sent again. After this finally verifies the order and supplies to the destination.
Getting a message from its source to its destination is fairly painless to understand. The internet is a store and forward network, meaning that those packets can be sent to any number of computers on their way to their destination. If there is a direct network link between two sites, that is, a physical cable linking the two computers the packets can zip right over.
Most of the time, there isn’t a direct link. Therefore, the sending computer sends the packets to one, that’s little closer to the destination. The machine moves the packets further down the line and so on, until the packets reach their goal. It’s not uncommon for a cross-country message to make 20 or 30 hops. Usually it all happens quickly. Open a telnet connection from Boston to New York or London and you’ll hardly notice any delay.