Using Temporary Flooding to Optimize Failover Times

MVRP advertisements use the active topology, which may be controlled through loop avoidance mechanisms like MSTP. When the active topology changes as a result of network failures, the time it takes for MVRP to bring up the optimal service connectivity may be added on top of the regular MSTP convergence time. Full connectivity also depends on the time it takes for the system to complete flushing of bad MAC entries.

To minimize the effects of MAC flushing and MVRP convergence, a temporary flooding behavior is implemented. When enabled, the temporary flooding eliminates the time it takes to flush the MAC tables. In the initial implementation, the temporary flooding is initiated only on reception of an STP TCN.

While temporary flooding is active, all the frames received in the extended data VPLS context are flooded while the MAC flush and MVRP convergence take place. The extended data VPLS context comprises all instantiated trunk SAPs regardless of the MVRP activation status. A timer option is also available to configure a fixed period of time, in seconds, during which all traffic is flooded (BUM or known unicast). When the flood-time expires, traffic is delivered according to the regular FDB content. The timer value should be configured to allow auxiliary processes like MAC flush and MVRP to converge. The temporary flooding behavior applies to all VPLS types. MAC learning continues during temporary flooding. Temporary flooding behavior is enabled using the temp-flooding command under config>service>vpls or config>service>template>vpls-template contexts and is supported in VPLS regardless of whether MVRP is enabled.

For temporary flooding in VPLS, the following rules apply.