Root Bridge Placement

Although STP is wonderfully automatic with its default values and election processes, the resulting tree structure might perform quite differently than expected. The Root Bridge election is based on the idea that one switch is chosen as a common reference point, and all other switches choose ports that have the best-cost path to the root. The Root Bridge election is also based on the idea that the Root Bridge can become a central hub that interconnects other legs of the network. Therefore, the Root Bridge can be faced with heavy switching loads in its central location.

If the Root Bridge election is left to its default state, several things can occur to result in a poor choice. For example, the slowest switch (or bridge) could be elected as the Root Bridge. If heavy traffic loads are expected to pass through the Root Bridge, the slowest switch is not the ideal candidate. Recall that the only criteria for Root Bridge election is that the switch must have the lowest Bridge ID (bridge priority and MAC address), which is not necessarily the best choice to ensure optimal performance. If the slowest switch has the same bridge priority as the others and has the lowest MAC address, the slowest switch will be chosen as the Root.

A second factor to consider relates to redundancy. If all switches are left at their default states, only one Root Bridge is elected, with no clear choice for a backup. What happens if that switch fails? Another Root Bridge election occurs, but again, the choice might not be the ideal switch or the ideal location.

The final consideration is the location of the Root Bridge switch. As before, an election with default switch values could place the Root Bridge in an unexpected location in the network. More important, an inefficient spanning-tree structure could result, causing traffic from a large portion of the network to take a long and winding path just to pass through the Root Bridge.

Figure 9-1 shows a portion of a real-world hierarchical campus network.

Figure 9-1 Campus Network with an Inefficient Root Bridge Election

Catalyst A 32768 00-00-00-00-00-0a

100Mbps Cost = 19

Catalyst C 32768 00-00-00-00-00-0c

Figure 9-1 Campus Network with an Inefficient Root Bridge Election

100Mbps Cost = 19

Catalyst C 32768 00-00-00-00-00-0c

1Gbps Cost = 4

Catalyst D

32768 Core Layer

00-00-00-00-00-0d

1Gbps Cost = 4

1Gbps Cost = 4

Catalyst E 32768 00-00-00-00-00-0e

Catalyst B 32768 00-00-00-00-00-0b

Access Layer

1Gbps Cost = 4

Catalyst D

32768 Core Layer

00-00-00-00-00-0d

1Gbps Cost = 4

1Gbps Cost = 4

Catalyst E 32768 00-00-00-00-00-0e

Server Farm

Catalyst switches A and B are two access-layer devices; Catalysts C and D form the core layer, and Catalyst E connects a server farm into the network core. Notice that most of the switches use redundant links to other layers of the hierarchy, as suggested in Chapter 2, "Modular Network

Design." At the time of this example, however, many switches, such as Catalyst B, still have only a single connection into the core. These switches are slated for an "upgrade," in which a redundant link will be added to the other half of the core.

As you will see, Catalyst A will become the Root Bridge because of its low MAC address. All switches have been left to their default STP states—the bridge priority of each is 32,768 (or 32,768 plus the VLAN ID, if the extended system ID is enabled). Figure 9-2 shows the converged state of STP. For the purposes of this discussion, the Root Ports and Designated Ports are simply shown on the network diagram. As an exercise, you should work through the spanning-tree process yourself, based on the information shown in the figure. The more examples you can work out by hand, the better you will understand the entire spanning-tree process.

Figure 9-2 Campus Network with STP Converged

Catalyst A 32768 00-00-00-00-00-0a

Catalyst C 32768 00-00-00-00-00-0c

Figure 9-2 Campus Network with STP Converged

Catalyst A 32768 00-00-00-00-00-0a

Catalyst C 32768 00-00-00-00-00-0c

Catalyst B 32768 00-00-00-00-00-0b

Catalyst E 32768 00-00-00-00-00-0e

Catalyst B 32768 00-00-00-00-00-0b

Access Layer

Catalyst D 00-00-00-00-00-0d

" Core Layer

Catalyst E 32768 00-00-00-00-00-0e

Server Farm

Notice that Catalyst A, one of the access-layer switches, has been elected the Root Bridge. Unfortunately, Catalyst A cannot take advantage of the 1-Gbps links, unlike the other switches.

Also note the location of the X symbols over the ports that are neither Root Ports nor Designated Ports. These ports will enter the Blocking state, and no data packets will pass through them.

Finally, Figure 9-3 shows the same network with the blocking links removed. Now you can see the true structure of the final spanning tree.

Figure 9-3 Final Spanning-Tree Structure for the Campus Network

Catalyst A 32768 00-00-00-00-00-0a

100Mbps

Catalyst C 32768 00-00-00-00-00-0c

1Gbps

Figure 9-3 Final Spanning-Tree Structure for the Campus Network

Catalyst A 32768 00-00-00-00-00-0a

100Mbps

Catalyst C 32768 00-00-00-00-00-0c

1Gbps

Catalyst B 32768 00-00-00-00-00-0b

Catalyst D

32768 Core Layer

00-00-00-00-00-0d

Catalyst E 32768 00-00-00-00-00-0e

Catalyst B 32768 00-00-00-00-00-0b

Access Layer

Catalyst D

32768 Core Layer

00-00-00-00-00-0d

Catalyst E 32768 00-00-00-00-00-0e

Server Farm

Catalyst A, an access-layer switch, is the Root Bridge. Workstations on Catalyst A can reach servers on Catalyst E by crossing through the core layer (Catalyst C), as expected. However, notice what has happened to the other access-layer switch, Catalyst B. Workstations on this switch must cross into the core layer (Catalyst D), back into the access layer (Catalyst A), back through the core (Catalyst C), and finally to the server farm (Catalyst E).

This action is obviously inefficient. For one thing, Catalyst A is probably not a high-end switch because it is used in the access layer. However, the biggest issue is that other access-layer areas are forced to thread through the relatively slow uplinks on Catalyst A. This winding path will become a major bottleneck to the users.

Was this article helpful?

0 0

Post a comment