Switching - Problems that Occur in Redundant Switched Topologies

Issues in Switching

Although switches are said to be ?intelligent? they are not without their problems:

Broadcast Storms

When a unicast packets with an unknown destination MAC gets flooded to all ports AND two switches are connected with two links (for redundancy), for example:

This can cause frames to loop around, because each switch will flood the packet back out the opposite port back to where it came from.

This is not good because Ethernet frames do not contain a TTL (Time To Live) field, the frames would therefore build up and go round and round forever. As the looping frames build up the performance of the switch will decrease and eventually result in a crash, typical behavior will be as follows:

The switch will work but will degrade in performance until it eventually gets too much and the switch crashes, it will then reboot and performance will be OK for a while until it starts to degrade and crash again.

• Because of these same loops as discussed above multiple copies of the same frame may be delivered to a destination.

• The MAC address table could get confused when multiple copies of the same frame are received on different ports. Frame forwarding could get impaired and CPU resources tied up in constantly updating the CAM table.