The TCP/IP suite (like its counterpart, the OSI model) uses the principal of a "stack". Each level of the protocol stack solves some set of problems, and has a "payload" area. The "payload" (data) contains data in another higher-level protocol, which solves a different set of problems. And so on.
The TCP/IP model was designed as the solution to a practical engineering problem. The OSI Model, on the other hand, was a more theoretical approach, and was built by committee. Therefore, the OSI Model is easier to understand, but the TCP/IP model is more practical. It is helpful to have an understanding of the OSI Model before learning TCP/IP, as the same principles apply, but are easier to understand in the OSI Model.
The TCP/IP Stack
There is some argument as to where the distinctions between layers are drawn; there is no one correct answer. But roughly:
+- - - - - - -+ 5| Application | i.e. HTTP, SMTP, FTP, TELNET + - - - - - - + 4| Transport | i.e. TCP, UDP +- - - - - - -+ 3| Network | i.e. IP + - - - - - - + 2| Data Link | i.e. Ethernet, PPP +- - - - - - -+ 1| Physical | i.e. Electricity, Radio Waves + - - - - - - +
The Physical Layer
This describes the physical characteristics of the communication, things like "binary 1 is +5V, binary 0 is -5V" (although, generally much more complicated.)
The Data-Link Layer
This specifies how packets are transported across a single wire. Ethernet, for example, includes details about specifying which machine on the network a packet is destined for. Some examples are Ethernet, PPP, Wireless Ethernet, SLIP and Token Ring.
The Network Layer
The Network Layer solves the problem of getting from the source network to the destination network. In the case of TCP/IP, this may involve routing the packet across a network of networks, known as an Internet.
IP contains information such as an address for the source and destination (in the form of IP addresses). It also solves other problems that arise from communication across distinct networks.
The Transport Layer
The protocols at this Layer can solve problems like reliability ("did the packet reach the destination?") and ensure that packets arrive in the correct order. It is also at this protocol where it is decided which application to connect to.
TCP is a very "sturdy" transport mechanism, which makes sure packets arrive in order, are re-transmitted if lost, and eliminates duplicates, as well as handling "emergency" content which must be handled out of order (out-of-band).
UDP is less sturdy. There is no attempt to verify that packets have reached their destination, and no guarantee that they will arrive in order. If the Application requires these guarantees, it must provide them itself, or use TCP. UDP is typically used for applications such as streaming media (audio and video, etc) where the time TCP requires for retransmission and re-ordering might not be available.
ICMP can be classified as either part of the Network or Transport layers, but doesn't really belong to either; it could perhaps be called layer 3 1/2. ICMP deals with the network itself. It is used to communicate facts such as "this link is congested; slow down" or "Sorry. Not available. Go away". The much-maligned Ping is a pair of ICMP packets.
The Application Layer
The application layer is where most common network programs reside. These programs and their corresponding protocols include HTTP (The World Wide Web), FTP (File Transport), SMTP (Email), TELNET (Remote Login), DNS (Name<-->IP Lookups) and many others.