Switching

Overview

Before investigating network protocols, we need to look at the major functions that a network protocol is responsible for - switching and routing. These two operations are closely related and often confused. Switching is the process of moving information through a network according to some path, while routing is the process of determining that path in an attempt to optimize the performance of the network. Different switching methods result in radically different network types and are critical in determining the properties of a network.

Why?

Every protocol has some set of functions that it performs on behalf of the layers above it. For a network layer implementation, what types of functions would you expect it to provide?
How does a packet of information from your computer get delivered to its destination through a sequence of network and routers or switches?
What the choices we have in deciding the best switching method to move data?
How do we evaluate the alternatives when making a choice in switching?

Objectives

Understand the term switching as it applies to computer networks.
Understand the difference between routing and switching.
Know the differences between circuit switching, packet switching and virtual circuit switching.
Understand cell switching as a technique for switching.
Understand the characteristics of ATM as a switching method.

Preparation

Read sections 3.1 and 3.3 and the notes.

Knowledge, Comprehension and Problem Solving

Terms

Asynchronous Transfer Mode (ATM)
ATM Adaption Layer
Circuit switching
Convergence Layer
Datagram
Packet switching
Routing
Routing table
Router
Segmentation and Reassembly Layer
Source Routing
Switching
Tunneling
Virtual circuit switching

Questions

Give an example of an application where circuit-switching would be appropriate and one where packet-switching is appropriate.

Given the following diagram of a virtual-switched network,

Show the virtual circuit tables as they develop for the following connections assuming that the virtual circuit identifier is always chosen as the lowest unused value.

Host 1 connects to host 3

Host 1 connect to host 6

Host 5 connects to host 6

Host 4 connects to host 3

Host 4 connects to host 2

Host 3 connects to host 5

For the same network, show a routing table for router C.
What is the maximum utilization for an ATM channel?
The IP datagram for a TCP ACK message is 40 bytes long: it contains 20 bytes of TCP header and 20 bytes of IP header. Assume this ACK is traversing an ATM network that uses AAL5 to encapsulate the IP packets. How many ATM packets are required to carry the ACK? How many if AAL3/4 is used?
Why are the ATM adaption layers provided?

Analysis, Synthesis and Evaluation

Write the datagram-handling algorithm for a router in a circuit-switched network? Do the same for a router in a packet-switched network?
Look at the VPI and VCI fields in the ATM header. Formulate a strategy for using these to perform clever switching.
Referring to question 5 above: IP and ATM are both Network Layer protocols; why are IP packets going across an ATM network and how does it happen?
Why might it be a problem to always start numbering virtual circuits at zero at a given switch?
Why have most networks moved to packet or cell switching as a mechanism rather than circuit switching? Why are many packet-switching networks trying to figure out some way to implement a form of circuit switching?
Explain why it would be impossible to have a virtual circuit scheme where the virtual circuit identifier would be constant as the packet moves through the network.
Look up Multi-Protocol Label Switching (MPLS) on the Internet (www.iec.org/online/tutorials/mpls is a good place to look) and compare it to the methods discussed here.