1. Switching When we have a lot of devices that want to communicate, how do we set up connections between them?

2. How to create point-to-point connections? On small networks, it’s okay to have direct connections between all devices (mesh topology) or connections to a central device (star topology). That gets obviously unwieldly and expensive rather quickly. Switching is the mechanism by which we create these connections with limited physical links available.

3. Circuit-switching In a circuit switched network, a connection between two devices is made through a series of physical links connected by switches. When the connection is created, a set of dedicated channels is set aside on each link so that a dedicated connection is formed.

4. Complications What complications will arise when we add switches to the network? –How do we pick one pathway from the many available? –How do we communicate to the sender and receiver what that pathway is? –How do we reserve the pathway while it’s being used? –How do we free the pathway when we’re finished with it?

5. Phases of communication Setup The sender, S, sends a request to the nearest switch with the address of the receiver, R. That switch finds a dedicated channel between itself and the next switch in the sequence, and forwards the request through that channel. The next switch creates a dedicated channel between itself and the next switch down the line….. etc., until the switch that R is connected to is found. R sends an acknowledgement back to S Once S receives the acknowledgement, a connection is established and S can start sending data. Data transfer Teardown: A signal is sent to each switch freeing up the channels

6. Packet switching Reserving a dedicated link between two devices is wasteful, especially if they don’t release the link when they’re not using it. Packet switching allows for maximal utilization of the physical links available on a network. This type of switching is typically software based, and done at the network level (so we’ll talk about it more there).

7. Virtual-circuit switching An attempt to blend the speed of circuit switching with the efficiency of packet switching As in circuit-switching, a unique pathway through the network is established. However, instead of this pathway consisting of dedicated channels, it consists of dedicated routing. So, whereas in a packet switched network, different packets can take different paths between S and R, in a virtual-circuit network, all packets between S and R take the same path.

8. Setup request phase: –In the setup phase, switches operate essentially as packet switches. –A request frame for a connection to R is sent from S to the first switch containing the virtual- circuit ID that S is assigning to this circuit. –The routing table says that packets from S to R go out through a particular port, so an entry in the virtual-circuit table is created. –Same thing happens at subsequent switches until the request reaches R, at which point R assigns a virtual-circuit ID from it’s end.

10. Acknowledgement phase –In order to complete the setup, R must acknowledge the request from S and communicate its VCI. –R sends an acknowledgement frame to its switch with the VCI it assigned to its virtual circuit with S. –That switch sends the packet to the previous switch, etc. until the acknowledgement gets back to S.