All information is transmitted across the Internet in small units of data called packets. Software on the sending computer divides a large document into many packets for transmission; software on the receiving computer regroups incoming packets into the original document. Similar to a postcard, each packet has two parts: a packet header specifying the computer to which the packet should be delivered, and a packet payload containing the data being sent. The header also specifies how the data in the packet should be combined with the data in other packets by recording which piece of a document is contained in the packet.
A series of rules known as computer communication protocols specify how packet headers are formed and how packets are processed. The set of protocols used for the Internet is named TCP/IP after the two most important protocols in the set: the Transmission Control Protocol and the Internet Protocol. Hardware devices that connect networks in the Internet are called IP routers because they follow the IP protocol when forwarding packets. A router examines the header in each packet that arrives to determine the packetís destination. The router either delivers the packet to the destination computer across a local network or forwards the packet to another router that is closer to the final destination. Thus, a packet travels from router to router as it passes through the Internet.
TCP/IP protocols enable the Internet to automatically detect and correct transmission problems. For example, if any network or device malfunctions, protocols detect the failure and automatically find an alternative path for packets to avoid the malfunction. Protocol software also ensures that data arrives complete and intact. If any packets are missing or damaged, protocol software on the receiving computer requests that the source resend them. Only when the data has arrived correctly does the protocol software make it available to the receiving application program, and therefore to the user