Per−VLAN Spanning Tree

You can have many instances of STP running in your network. By running a different instance of STP on a per−VLAN basis, you can run some VLANs on ports that are blocked by another instance of STP running on another VLAN. In this way, you can set the priority of each port on a per−VLAN basis, allowing you to use the redundant links in your network to run an equal amount of traffic on each link. The VLANs individually determine which links to forward and which links to block.

Just as with the port priority setting, the port with the lowest priority value for each VLAN gets to forward the frames. If more than one or all the ports have the same priority value for a particular VLAN, the port with the lowest port number gets to forward the frames for that VLAN.

PVST and PVST+

Per−VLAN Spanning Tree (PVST) is a Cisco proprietary solution to the scaling and stability problems associated with Common Spanning Tree (CST) in large−scale spanning tree networks. PVST creates a separate instance of STP on each VLAN in the switch block. This setup gives each VLAN a unique STP topology containing its own port cost, path cost, priority, and root switch.

By using separate instances of PVST on each VLAN, you reduce the convergence time for STP recalculation and increase reliability of the network. By implementing PVST, the overall size of the spanning tree topology is reduced significantly. PVST improves scalability and decreases convergence time, providing faster recovery in the event of network faults. It also allows control of forwarding paths on a per−subnet basis while providing a simple technique for Layer 2 redundancy.

PVST does have some disadvantages in the spanning tree. PVST uses more processing power and consumes more bandwidth to support spanning tree maintenance and BPDUs for each VLAN. Inter−Switch Link (ISL) uses one spanning tree per VLAN, using PVST over ISL trunks. PVST implementation requires the use of Cisco ISL encapsulation in order to function.

Per−VLAN Spanning Tree Plus (PVST+) is not well documented by Cisco. IEEE 802.1Q can use PVST+ to map multiple spanning trees to the spanning tree of authentic IEEE 802.1Q switches.

PVST+ is available in Catalyst software versions 4.1 or newer. Cisco Catalyst switches configured with version 4.1 or later are considered Cisco PVST+ switches. PVST+ is compatible and interoperable with legacy type Mono Spanning Tree (MST) and PVST switches without any user intervention—thus PVST+ has a type of plug−and−play functionality.

Configuring the port priority by VLAN is useful for distributing data across parallel paths. If a parallel connection exists between two devices, STP will block one of the links. If the port priority is not changed on one of the ports for each VLAN, the traffic from all VLANs will travel on one link, and one link will be used only as a backup.

When you need to have multiple links to a destination be able to transmit data as though they were one link, you can use EtherChannel. Let’s take a look at EtherChannel in the next section.

EtherChannel

Fast EtherChannel can provide scaled bandwidth within the campus using full−duplex bandwidth at wire speed for up to eight bundled links. A bundle is a series of links acting like a single link between two points in

209

the network. Thus, eight wires can be used to simulate one link able to handle up to 800Mbps and load balance data across those links, as shown in Figure 10.7.

Figure 10.7: Eight equal−cost links between two switches, creating a bundle of eight channels acting as a single link.

Let’s take a look at what occurs during a link failure in an EtherChannel bundle. We’ll also examine the Port Aggregation Protocol (PAgP).

Link Failure

Fast EtherChannel provides redundancy in the event of a link failure. The EtherChannel bundle is managed by the Fast EtherChannel process and the Ethernet Bundle Controller (EBC). Should one link in the bundle fail, the EBC informs the Enhanced Address Recognition Logic (EARL) ASIC of the failure. The EARL ASIC immediately ages out all addresses learned on that link. The EBC and the EARL then recalculate in the hardware, sending queries to the other switches and learning the destination link based on the responses. The data traffic is rerouted on one of the other links in just a few milliseconds, making the convergence transparent to the user.

EtherChannel Administrative Groups

You can define an EtherChannel administrative group to identify groups of ports that are allowed to form an EtherChannel bundle together. When you create an EtherChannel port bundle, an administrative group is defined automatically. Administrative group membership is limited by hardware restrictions.

The administrative group can be any value between 1 and 1,024, inclusive. It is defined using the

set port channel <port list> <administrative group number>

command. To view the configured administrative groups, use this command:

show channel group <administrative group number>

Modifying a member port of an EtherChannel administrative group will cause the port to be removed from the group when STP realizes that a change has occurred in the network topology. The modified port that was a member of the EtherChannel administrative group must go through listening and learning mode again before it can return to forwarding mode and rejoin the EtherChannel bundle.

Port Aggregation Protocol

The Port Aggregation Protocol (PAgP) is used to manage the Fast EtherChannel bundles and aids in the automatic creation of Fast EtherChannel links. PAgP packets are sent between Fast EtherChannel−capable ports. PAgP learns of the neighbors and their group capabilities dynamically and then informs its neighbors of the local group capabilities. After the protocol determines all the paired, point−to−point, or bi−directional links, it groups into a single channel those ports that have the same neighbor device ID and neighbor group

210