Overview

The Internet Protocol (IP) may best be described as a connectionless datagram delivery system. It is connectionless since each unit (a datagram) is delivered in isolation, even with reference to other related packets. With IP there are no connections nor logical circuits. Many say that the service provided is unreliable since no guarantees are made that the datagram will ever arrive at its destination. IP provides no acknowledgements; datagrams may be lost, duplicated, or arrive out of sequence at the destination host. IP has no way in which to inform the transmitting station. Equally, IP has been described as being a best efforts delivery system. This is because datagrams may legitimately be discarded due to insufficient resources, again without informing the source host. While true, these statements paint a bleak and unfair picture of what has become an extremely robust and versatile protocol.

IP, as outlined by RFC 791, defines that IP performs two basic functions, namely addressing and fragmentation. In the previous chapter we saw the basic addressing scheme employed and how this is split into two sections, specifically the Network ID and the Host ID. Also in the previous chapter we saw how routers or Gateways are used to connect different networks. In Chapter 9, we shall see how this routing works in detail, by being able to determine paths to route datagrams towards their eventual destinations. In this chapter we shall see that, through the use of specific fields in the IP...