11.6 Other Network Layer Services 11.6.1 Connectionless network services Instructor Note The concept of connectionless network services is introduced. This is a fundamental property of the Internet -- packets can take various paths to get to their destination. This helps ensure delivery if one path becomes unavailable for some reason. Using the diagram, simply erase one of the links or one of routers and have the students note that multiple (redundant) paths to a destination is a very desirable feature of an internetwork. The students should be reminded of the postal system analogy, where zip codes are like IP addresses and where the post office performs the routing functions. Most network services use a connectionless delivery system. They treat each packet separately, and send it on its way through the network. The packets may take different paths to get through the network, but are reassembled when they arrive at the destination. In a connectionless system the destination is not contacted before a packet is sent. A good analogy for a connectionless system is a postal system. The recipient is not contacted before a letter is sent from one destination to another. The letter is sent on its way, and the recipient learns of the letter when it arrives. 11.6 Other Network Layer Services 11.6.2 Connection-oriented network services Instructor Note The concept of connection-oriented network services is introduced. The telephone system, which relies on connection -- real physical circuits between source and destination -- is given as an example. Some data networking technologies are connection-oriented, but they will not be focused upon until later semesters. Point out to students a potential flaw in connection-oriented systems -- if at any point the circuit is disrupted, the communication stops. In connection-oriented systems, a connection is established between the sender and the recipient before any data is transferred. An example of a connection-oriented network is the telephone system. You place a call, a connection is established, and then communication occurs. 11.6 Other Network Layer Services 11.6.3 Comparing connectionless and connection-oriented network processes Instructor Note This target indicator identifies another contrast between connectionless and connection-oriented network processes -- information can arrive out of order in a connectionless system, whereas information arrives sequentially in a connection-oriented system. Thus connectionless systems must have some provision for correctly ordering data as it arrives at the destination host. Connectionless network processes are often referred to as packet switched. In these processes, as the packets pass from source to destination, they can switch to different paths, as well as (possibly) arrive out of order. Devices make the path determination for each packet based on a variety of criteria. Some of the criteria (e.g. available bandwidth) may differ from packet to packet. Connection-oriented network processes are often referred to as circuit switched. These processes establish a connection with the recipient, first, and then begin the data transfer. All packets travel sequentially across the same physical circuit, or more commonly, across the same virtual circuit. The Internet is one huge connectionless network in which all packet deliveries are handled by IP. TCP (Layer 4) adds connection-oriented services on top of IP (Layer 3). TCP segments are encapsulated into IP packets for transport across the Internet. TCP provides connection-oriented session services to reliably deliver data. 11.6 Other Network Layer Services 11.6.4 IP and the transport layer Instructor Note IP is identified as a connectionless network service. This has as its historical roots the fact the Department of Defense wanted a network that could survive a war which destroyed parts of the network. For such a network to ensure that messages could still be delivered as parts of the network were being destroyed, the concept of packet-switching and the specific implementation of IP were developed. IP is a connectionless system; it treats each packet independently. For example, if you use an FTP program to download a file, IP does not send the file in one long stream of data. It treats each packet independently. Each packet can travel different paths. Some may even get lost. IP relies on the transport layer protocol to determine whether packets have been lost, and to request retransmission. The transport layer is also responsible for reordering the packets.
