Brocade Network OS Administrator’s Guide v4.1.1 - Manual (Page 389)

‐ The VCS Fabric and the attached vDCs belong to the same MSTP region.
‐ VLAN-to-instance mapping must be the same in the VCS Fabric and for each vDC.
‐ An MSTP instance topology is formed by the VCS Fabric (which appears as a single logical
switch) and all attached vDCs.
‐ Service VF configuration is allowed. Service VFs (VLAN IDs greater than 4095) are not
assigned to any instance and are always in a forwarding state.
‐ A topology change in one MSTP instance for a vDC affects the topology of the same
instance in other vDCs.
‐ Data loops between the VCS Fabric and individual vDCs are detected by each MSTP
topology.
STP participation
The default state for all VLANs (802.1Q or service VFs) is "no spanning-tree shutdown."
For RSTP, there is one RSTP instance in the VCS fabric; the protocol is still VLAN-unaware. This RSTP
instance consists of switch ports that have identical VLAN configurations. The VLAN configuration may
consist of 802.1Q VLANs or classified VLANs. The STP port state applies to all VLANs (802.1Q and
classified) on a port. For switch ports that cannot participate in the same RSTP instance, it is the user’s
responsibility to shut down spanning tree on these ports. This is the case where ports have overlapping
C-TAG classifications and these C-TAGs represent different service VFs.
For MSTP, the VCS Fabric is a single MSTP region. The VLAN-to-instance mapping is applicable only
to 802.1Q VLANs. The MSTP VLAN digest calculation is based on 802.1Q VLANs alone. The MSTP
state is applied on a port-instance basis (as in previous releases). For switch ports that cannot
participate in the same MSTP instance, it is the user’s responsibility to shut down spanning tree on
these ports. This is the case where ports have overlapping C-TAG classifications.
Service VFs can participate only in MST instance 0 and cannot be assigned to another MST instance.
When a service VF is shut down, it is assigned to an internal instance (instance 255) that is always in
the forwarding state. The default state for all 802.1Q VLAN and service VFs is “no spanning-tree
shutdown.”
For PVST, the STP instance is on a per-service-VF basis. PVST can be enabled only on a service VF
that has a classification tag. The classification tag identifies the default 802.1Q VLAN in the attached
network and is carried in the PVST BPDU; it is also used to form the root RBridge ID. The service VF
must have the same C-TAG classification and a nonconflicting classification with other VFs on all
RBridge interfaces. This is to ensure the uniqueness of the root RBridge ID for each PVST instance.
Consequently, note the following conditions:
• PVST cannot be enabled on a service VF with a VLAN ID greater than 4095 on an access port.
• PVST cannot be enabled on a trunk-mode native VLAN that has no C-TAG classification.
For edge-loop detection (ELD), the protocol instance is on a per-service-VF basis. The user can use the
CLI to enable ELD for any service VF on a switch port. Because an ELD configuration applies on a port,
the classification for that service VF must exist before the ELD configuration can be accepted.
STP tunneling
BPDU tunneling can be controlled on a per-port basis by means of the existing bpdu-drop command.
In a multitenancy environment, where a VCS Fabric could be connected to multiple STP-enabled
networks, the fabric should tunnel only one instance of the STP BPDU. It is the user’s responsibility to
enable tunneling on ports that belong to the same STP instance. Other switch ports should have
tunneling disabled.
Currently, the bpdu drop command controls BPDU forwarding only on the ingress port; the forwarding
decision must be applied on the egress port as well. Nontagged BPDUs are tunneled on VCS control
VLAN 4095. At the egress RBridge, switch ports that have BPDU tunnels disabled should be removed
STP participation
Network OS Administrator’s Guide 389
53-1003225-04