Circuit Switching and Packet Switching

• Look at the pictures on page 311, 312
• In a circuit switched network
  • A dedicated circuit is established at the beginning of the connection
  • Resources are allocated for this circuit for the duration of the connection.
  • All data flows over this circuit
  • At the end of the connection, the circuit is deallocated.
• In a packet switched network
  • Individual packets are sent along the current "best" path.
  • Resources are not preallocated.
  • The network may change and the data transmission will adapt.
  • This generally involves more overhead for each packet.
• A communications network
  • Consists of a number of switching nodes
  • This can be viewed as a graph.
  • The nodes are the vertexes
  • The edges are the communications links.
  • These are usually connected, but not fully connected.
  • There are usually multiple paths through the network.
  • The object is to route data through this network from one end station to another.
• Circuit Switching
  • Old style operator actually making a physical connection.
  • Typical Session
    • Establish a circuit through the network and allocate resources
    • Transfer data along the circuit
    • Disconnect, disassemble the circuit
  • The circuit is allocated when established
    • And stays allocated until the circuit is disconnected.
    • This means that there is little delay through the circuit after it is established.
    • And that there is sufficient network capacity to handle whatever traffic it allocates circuits for.
    • And that the resources are not available to other connections.
    • This is great for a phone call
    • This is not so good for a telnet session.
  • A switch
    • Contains a CPU, and network interfaces.
    • Consists of a number of switching elements connected in a switched fabric.
    • 
    • A switch element can connect any two items.
    • 
    • Note that when the switch is "full" only a small number of the switching elements are actually used.
    • 
    • And as the size of the switch grows, the number of switching elements grows by n²
    • And a single element failure is critical. (Why?)
    • But this switching fabric is non-blocking
  • Multiple stage switches are constructed.
    • To reduce the number of SE'
    • To provide redundancy in the face of failure.
    • But they become blocking.
    • 
  • These are examples of space division switching.
  • Time division switching also used, but Stallings tells us it is outside the scope of the book.