Lecture 6 (IP)

Slide 1 [Introduction]
In today's and next week's lecture we will cover two of the most important areas in networking and the Internet: IP and TCP. These cover the network and transport layer of the OSI model, and we will generally be investigating the functions of these layers, and how IP and TCP fit into them.

The Internet was designed by DARPA, and its main purpose was to create a robust network which could withstand a strike on any part of the network. It was created using the IP protocol, which defines IP addresses for each of the nodes on the network. These addresses have to be unique over the whole of the Internet. Its greatest problem since has been its success, as we are now running out of IP addresses to assign to all the devices which want to connect to the network.

You should not really think of the Internet as just being a collection of desktop computers. The future will move away from these large desktop systems, towards small, mobile devices. It is predicted that within a couple of years that there will be more mobile devices which connect to the Internet, than static desktop systems. In order to cope with this demand we need many more IP addresses. At present most systems use IP Version 4, which has a 32-bit network address, giving around 4,294,967,296 network addresses. The new standard (IP Version 6) will have 128 bits for the network address, which gives over 340 million million million million million million address (3.4x1038), which should be enough to go around, for a while. Just think of the future? Every car could be truly Internet-enabled, where each embedded computer system (such as the engine management system, or the air conditioning system) could be uniquely addressable over the Internet.

TCP and IP are two of the most important protocols ever developed, and have done more for world peace and any diplomat has ever done. In fact, TCP and IP unite the world and allow everyone in the world to communicate, no matter which computer they use, which operating system they are running, which language they speak, or which network they use.

So what really makes the Internet work? Why does the WWW work so well? How can we run so many applications over the Internet at the same time? How do we know that our data has been received? How does the data actually know how to get to a certain destination? Well, it's to do with TCP, IP and routing protocols. These three parts make the whole of the Internet work, and work reliably. The IP part is responsible for getting the data packets from the source to the destination (using the IP address), the TCP part is then responsible for sending the data to the required application program (using TCP sockets and sequence numbers), and the routing protocols are responsible for passing on information about how to get to destinations (using protocols such as RIP). Isn't it wonderful how a user can run a few WWW browsers, a TELNET session, an FTP session, a video conferencing session, and it all works, seamlessly, even if there are multiple destinations.

As we'll find out, in most cases we need TCP to go along with IP, as IP is only responsible for identify the destination and the sending address, and deciding if the data on the network is for them. It is TCP which either tags all the outgoing data and clearly identifies the virtual connection, or reorders and passes the received data to the required application. It also makes sure that all the data is properly received, and that anything that is sent, that there must be receipt it. If a receipt is not received, the data is re-sent.

Slide 2 [Data encapsulation]
This module is based on the 7-layer OSI model, and each unit focuses on a particular layer of the model. With Ethernet and ATM were are looking at Layer 2 (the data link layer), in this unit we are looking at the network layer, and in the next unit we will look at the transport layer. This unit will provide us with a greater understanding of how the Internet works, as the Internet is basically just an infrastructure of networks which use a standard addressing structure: IP addresses.

To simply the process of sending and receiving data, the network uses a layered approach, where each layer encapsulates the data with information that is required at the given layer. At the presentation layer we would define things like the character set that we were using (such as using an ASCII character set), the type of encryption, and so on. Next at the session layer we would build in data which allowed us to create and maintain a session on the remote system. This might involve us defining our user name and password, or some way to define the destination of our data.

At the transport layer a great deal of processing occurs. It is responsible for many things, including:

- Splitting the data up into data segments, which are small enough to be able to be sent over the network connection. As we will see Ethernet only allows for a maximum of around 1500 bytes of data to be sent, in an Ethernet frame.
- Identifying each data segment, so that missing segments can be identified, and that out-of-sequence data segments can be rearranged into the correct order.
- Making a connection with a remote system, and acknowledging the data that has been sent.

and several other functions (that will be discussed next week).

The network layer is mainly responsible for network addresses, and routing the data around the Internet, and, at the sending end, basically just adds the network addresses of the source and destination onto the data segments produced from at the transport layer. Eventually the data link layer takes the data packet produced by the network layer and formats it in a way which can be transmitted over the local network. This normally involves addressing the physical source address of the local system, and the destination address, which is the next device to receive the data frame within the local network.

Slide 3 [IP and TCP]
IP and TCP fit into the network and transport layers of the OSI model, respectively. The main purpose of the network layer is to define network addresses, and to determine the best path to a destination. The physical address of a node only has local significance, whereas the IP address has significance over the whole of the Internet (or an internet, if it is a locally defined network). Devices called routers, read the destination IP address, and make a decision as to whether they should forward the data packet onto another network. This decision occurs at the network layer, thus routers only require to implement the first three layers of the OSI model (although, for security reasons, they may also read transport layer information).

Once the data packet is delivered at the destination, the function of the network layer is complete. It does not even ask for a receipt that the data packet has been received correctly. The transport layer then takes over and streams the data segments to the required application stream.

For the rest of this lecture, see the PDF version.