STP-BPDU TYPES

STP Topology Changes:

Switches exchange two types of BPDUs when building and converging the topology database:

• Configuration BPDUs

• Topology Change Notification (TCN) BPDUs

Configuration BPDUs are used to elect Root Bridges, root ports, and designated ports.

A TCN will be sent under two circumstances:

• When a port transitions into a forwarding state.

• When a forwarding or learning port transitions into a blocking or down state.

When a topology change occurs, a switch will send a TCN BPDU out its root port, destined for the Root Bridge. The TCN contains no information about the change – it only indicates that a change occurred.

Consider the following example:

If the port on SwitchD connecting to SwitchE went down:

• SwitchD would send a TCN out its root port to SwitchB.

• SwitchB will acknowledge this TCN, by responding with a TCN with the Topology Change Acknowledgement (TCA)   flag set.

• SwitchB then forward the TCN out its root port to SwitchA, the Root Bridge.

Once the Root Bridge receives the TCN, it will send out a configuration

BPDU to all switches, with the Topology Change (TC) flag set. This ensures that all switches in the STP topology are informed of the change.

When a switch receives this root BPDU, it will temporarily reduce its CAM aging timer from 300 seconds to a value equal to the forward delay timer -15 seconds by default. This allows any erroneous MAC addresses to be quickly flushed from the CAM table.The CAM aging timer will remain at a reduced value for the duration of one forward delay plus one max age period – a total of 35 seconds by default.

Two types of failures can occur in the STP topology, depending on the perspective of a switch:

• Direct failures

• Indirect failures

For example, if the root port on SwitchB fails, SwitchB would consider this a direct failure. SwitchB will detect immediately that the physical port is down, and STP will react accordingly.

That same port failing would represent an indirect failure for SwitchD. SwitchD would lose its path to the Root Bridge. However, because the port is not local on SwitchD, it must learn of the topology change from its neighbors.

By detecting and reacting to link failures, STP can take advantage of the

redundancy provided by loops. However, the failover is not instantaneous.

If the root port on SwitchE were to fail:

• SwitchE would immediately purge any BPDU information received from SwitchC.

• SwitchC would send a TCN to the Root Bridge.

• The Root Bridge would send a configuration BPDU to all switches,with the TC flag set.

• All switches would reduce their CAM aging timer to 15 seconds.

• SwitchE would eventually receive a BPDU from SwitchD.

Remember: blocked ports still receive BPDUs to learn about topology changes.

• The blocked port to SwitchD now represents the best and only path for SwitchE to reach the Root Bridge.

• The blocked port will transition first to a listening state, and then to a learning state. The port will wait the forward delay time in both states, for a total of 30 seconds.

• The port will finally transition to a forwarding state.

Thus, hosts on SwitchE will be impacted by this failure for a minimum of 30 seconds. STP will maintain redundancy if there is a loop, but a link failure will still negatively impact the network for a short period.

Refer to the below links for more detailed learning.
http://www.cisco.com/c/en/us/support/docs/lan-switching/spanning-tree-protocol/12013-17.html