Network managers designing Layer 2 networks have long relied on IEEE 802.1D Spanning Tree Protocol to provide redundancy while ensuring loop-free connectivity among multiple network bridges and switches. However, the combination of 802.1D and 802.1Q virtual LANs creates network-architecture challenges.
If you have multiple links to separate VLAN traffic, STP could disable some of those data paths. 802.1S Multiple Spanning Tree Protocol solves the problem by supporting multiple spanning trees within a network. The standard lets administrators assign VLAN traffic to unique paths.
Consider a network configuration of three switches that are fully interconnected. Within the network are two VLANs with IDs of 10 and 20. Switch 1 has VLAN 10 and 20 assigned to two unique ports on the switch so that VLAN 10 and VLAN 20 traffic flows over separate links. At first glance this appears to be an ideal configuration to load balance the traffic over the two VLANs (see graphic). However, STP is running on all three switches in this network.
With S3 chosen as the root bridge, STP will block the link between Switches 1 and 2. When this occurs, the traffic from VLAN 20 can't transverse the network. This problem arises because, while the switches treat VLAN 10 and 20 as complete separate networks, the original 802.1D-based STP treats the overall topology as a single network because it doesn't have the concept of multiple networks.
One solution would be to run multiple, independent copies of STP, known as a spanning-tree instance, on the switch. But assigning a unique spanning-tree instance to each VLAN isn't practical because this introduces overhead on the switches. What's more, most networks don't need more than a few logical topologies. Rather, one spanning-tree instance per desired topology should suffice.
For multiple devices to properly interact they must be aware of the mapping of VLANs to multiple-spanning-tree instances. In large enterprise networks there might be a need to have different VLAN-to-MSTP instances, and as such the 802.1S standard accommodates these different mappings through the use of multiple-spanning-tree regions.
Looking back at the initial example, you can see how the use of 802.1S solves the topology problem. If you assign VLAN 10 to MSTP Instance 1, and VLAN 20 to MSTP Instance 2, there will be two separate spanning-tree topologies. Switch 3 will become the root bridge for Instance 1 and will block the link between Switches 1 and 2.
But unlike the 802.1D-based scenario, this link is blocked only for traffic from VLAN 10. Traffic from VLAN 20 can traverse this link. Likewise, MSTP Instance 2 chooses Switch 2 as its root bridge and blocks the link from Switches 1 to 3 for traffic from VLAN 20.
By assigning VLANs to separate spanning-tree topologies, network managers ensure that both VLANs can traverse a network appropriately. This produces the desired effect of balancing traffic across the networks and reveals the value of 802.1S MSTP in a network topology.
Ward is program director of product management for LVL7 Systems. He can be reached at firstname.lastname@example.org.