Port States

 The operational states of ports participating in the STP are a bit different from a conventional switch. Additional states are needed to prevent a loop, and to limit instability during the voting or topology-change process. There are five states.
Disabled — This port is completely non-functional in that it cannot receive or transmit any type of frame.

Blocking — This port is neither a designated nor root port but is recognized as an alternate port to the root. It does not learn addresses, forward frames or transmit BPDUs. It can hear BPDUs being sent since it may be called to action one day.

Listening — This port is being prepared for activity by exiting the blocking state. It still does not learn or forward addresses, but it sends and receives BPDUs. It is participating in the voting, but it might not win the election.

Learning — This port will become active in forwarding frames but must wait until the Forward Delay timer expires (typically 15 seconds). This allows the port to add entries in its filtering database so it will not flood ports once it enters the forwarding state.

Forwarding — This port is functioning as any other switch port by filtering and forwarding frames.
We will go back to our example in Figure 2. Since we have a tree topology without loops, all active ports (those with a link partner) are in the forwarding state. Now consider the presence of Link U. Unused Port 4 on Bridges A and B will be active but only in the listening state. While in this state, they send each other BPDUs. Bridge A will send out a BPDU to Bridge B indicating a root path cost of 19 while Bridge B reciprocates with a root path cost of 100. Both bridges recognize there is an alternate path to the root, which is unacceptable, and that (for messages not originating at the end stations attached to Bridge B) the best path to the root is through Bridge A and not Bridge B. Therefore Port 4 on Bridge B assumes the blocking state, and Port 4 on Bridge A the learning state — and eventually the forwarding state, even though this would be useless since all the messages forwarded from Bridge A to Bridge B would be discarded on arrival. The final result is no topology change. The tree before the redundant connection is the same as after the connection was made. However, Link U is now a potential redundant path which may be utilized after a link or switch failure.

Topology Change

In the above example, adding Link U did not result in a change in the tree topology. However, a topology change can occur due to a lost link, a lost bridge, the addition of a link or bridge, or by management changing the priorities of bridges. What happens if the root bridge fails? STP guards against all these occurrences by monitoring configuration BPDUs, observing that a BPDU failed to arrive, or by generating a Topology Change BPDU.

There are two types of BPDUs as identified in the BPDU Type field. The configuration type is the normal BPDU as shown in Figure 3. The topology-change BPDU is similar to the configuration BPDU except that no data is transmitted below the Type field. This BPDU is generated by one of the designated bridges that changed its topology. An intervening designated bridge will acknowledge the originator's topology-change message by sending a configuration message with the Topology Acknowledge bit set in the Flags field. A new topology-change BPDU will then be sent towards the root. Any intervening designated bridge would repeat the process until the root is notified. The root bridge notices the message and informs all its attached designated bridges of the topology change by setting the Topology Change bit in the flags field and sending out a new configuration message. While this flag is set, all designated bridges reduce their aging time to that of the Forward Delay timer in anticipation of the topology change. Since the topology change could possibly make the data in the filtering database invalid, it must be quickly cleared and the new location of end stations relearned. Under normal conditions, the standard recommends a default aging time of 5 minutes! Changing the time to 15 seconds would be a great help in relearning address locations. Only after the root clears the topology change bit will the designated bridges resume their normal aging time and begin the learning and forwarding operations for the new topology.

SUMMARY

This article provides an introduction to STP. Since the protocol is quite complex, not all issues have been addressed. Since protocol timers and aging times can vary, it is impossible to predict the time it would take for a network to stabilize after a topology change. STP has been criticized for being too slow in implementing a topology change in an industrial network; however, the concepts are similar to faster redundancy schemes such as Rapid Spanning Tree Protocol—so STP should be learned. One advantage of STP is that it is not specific to Ethernet and can operate over wide area networks (WANs) as well. For supervisory control and data acquisition systems (SCADA), the speed of recovery to topology changes might be adequate when temporary loss of communication will not render local control useless.

References

ANSI/IEEE Std 802.1D, 1998 edition: Information technology—Telecommunications and information exchange between systems—Local and metropolitan area networks—common specifications—Part 3: Media Access Control (MAC) Bridges, The Institute of Electrical and Electronics Engineers, Inc.

The Switch Book, Rich Seifert, 2000 Wiley Computer Publishing.