Portfast and Multicast traffic
In my last post I covered the Spanning-Tree Portfast feature and the advantages it has on our network, covering the concept of TCN packets and the effect TCN packets have on our networks. I did however forget to mention another side effect the TCN packets cause involving multicast traffic. When a topology change occurs and a TCN packet is set to the spanning-tree root bridge the root bridge transmits a global leave message with a multicast group address of 0.0.0.0, this in turn causes the switches to flood multicast traffic out all ports until it receives a few general queries (the default number will depend on the switch model and IOS). So this simply identifies another potential resource issue that can be caused by not configuring Portfast. There a few configuration options we can set that can make this process less intrusive:
- Outright disable the flooding of multicast traffic when a TCN is transmitted, This is configured on an interface by interface basis.
- Tell the router to transmit a query message – Configured on the switch from global config mode, whenever a switch notices the TCN packet it sends a packet out to tell the router to solicit a query message, decreasing the amount of time it takes the switches to receive “x” amount of queries to recover from the TCN event.
- Specify the number of queries to wait before the switch stops flooding multicast out all ports – This configures the switch to stop flooding multicast traffic after it has received “x” amount of queries. This feature in conjunction with the solicited query message feature can significantly reduce the flooding time.