CCNP BCMSN Exam Cram 2 (Exam Cram 642-811)

This section discusses some basics about Layer 2 redundancy design methods and Cisco solutions. Because STP and its operation and enhancements were already discussed in Chapters 3 and 4, they are not mentioned here.

Uplink Interfaces

A switch's SE contains two Gigabit Ethernet uplink interfaces. Cisco assumes that you'll use these interfaces to connect to a switch at either a higher or lower layer in Cisco's three-layer hierarchy: core, distribution, and access. Of course, you can connect anything you want to these interfaces. However, you do have to take care of setting up redundant connections between a higher and lower layer.

For example, let's look at the right side of Figure 7.1, where there are redundant connections between a switch and its lower- or upper-layer neighbor. I'll focus on the Distribution MLS switches at the bottom of this diagram. Let's assume that the distribution layer is using Catalyst 6500 switches with dual SE cards: primary and secondary. The left side of Figure 7.2 shows a diagram of the connections on the Catalyst 6500 to the two switches at the core layer (the backbone switches). Notice that the primary SE's two Gigabit uplinks are used in this design. The main problem with this approach is that if the primary (top slot) SE fails and the secondary is promoted, all connectivity to the backbone is lost because both connections were connected to the SE in the top slot.

A better solution is shown in the right side of Figure 7.2. In this example, the two connections to the two core switches are placed on different cards (the two uplink ports on the SEs). Remember that you can do this with the secondary SE because the Gigabit ports are activated, even in a standby mode. In this example, if the primary SE fails in the top slot, the Gigabit uplink on the secondary SE is still functional.

Remember that this example applied to SEs, but you should follow the same practice for other line cards in your chassis. One uplink connection should be in one line card and a second connection should be in a different line card.

Switch Redundancy

In most situations, if you're really concerned about redundancy, you should use chassis redundancy rather than component redundancy. Chassis redundancy provides a more robust form of redundancy, but requires you to buy two switches at the distribution and core layers, which can be costly. The right side of Figure 7.1 shows an example of this approach.

Remember that in this design, the distribution layer contains a Layer 3 process that provides a boundary between the access layers and the core layer. This is used to contain Layer 2 problems, such as STP issues, broadcast storms, and so on. You must create your VLANs correctly, tune STP, set up a routing protocol, and possibly configure MLS to create a well-designed, highly optimized, redundant topology. All of these tasks were covered in chapters leading up to this one.