Introduction to Availability and Redundancy

There are two methods of providing basic redundancy services: chassis redundancy and component redundancy. Chassis redundancy has another networking device, with the same functionality, that provides redundant services in case the primary device fails. Component redundancy provides for redundant components inside the same chassis. Six key components are used to reach a high level of availability (as close to 100% as possible):

• Network device reliability Redundant hardware components and intelligent software should be able to deal with a failed component.

• Device redundancy Redundant devices are used in case one device fails.

• Link redundancy Backup network connections are used in case a primary network connection fails.

• Fast convergence Intelligent software is used to converge very quickly if a component, Layer 2, or Layer 3 failure occurs.

• Good network design For convergence to take place in an acceptable timeframe, a network design must exist to make use of a fast convergence solution, such as UplinkFast for STP or component redundancy in key networking devices.

• Documented network A well-documented network helps to pinpoint weaknesses so that corrective measures can be taken.

Of the preceding bullet points, the most important is the creation of a network design that minimizes any weak points by providing an appropriate level of redundancy. In other words, it is sometimes too costly, and unnecessary, to provide 100% redundancy. The important point is that you need to rank your networking resources from critical to nonessential, and then determine what level of redundancy is needed and what type of solutions you should use to provide that redundancy. The following two sections discuss the two approaches that can be used to provide redundancy.

Remember the six bullet points to provide a high level of availability.

Component Redundancy

One type of redundancy is component redundancy. This type of redundancy provides protection against component failures inside the same chassis. Components that can be protected in some Catalyst switches include Supervisor Engines, modules (hot-swappable), power supplies, and fans.

If you decide to implement only component redundancy (no backup devices), your network would look something like the one shown on the left side of Figure 7.1. Notice that there are no redundant devices in this example. However, there are redundant connections between the devices, which you can see in the figure, and redundant components within those devices.

Figure 7.1. Component versus chassis redundancy.

Using a component redundancy design provides these benefits:

• All single points of failure have been eliminated.

• The different switches do not have to be located in the same geographic area.

• Minimizes convergence because this process takes place within a device and not necessarily between devices.

However, there are downsides to using only component redundancy:

• Redundant components in a chassis are typically in a standby state and cannot be used unless the primary component fails.

• Because all components must be duplicated, the cost of the switch becomes very expensive.

• Component redundancy cannot deal with all types of failure, especially application layer processes, backplane failures, and complete device failures.

Chassis Redundancy

Chassis redundancy provides redundant networking devices. An example of this is shown on the right side of Figure 7.1. In this example, there are redundant switches at each layer of the design. Enhanced STP features are used to provide quick STP convergence in case of a Layer 2 failure, and an intelligent routing protocol such as OSPF or EIGRP provides quick Layer 3 convergence.

This type of design provides the following advantages:

• You do not need to provide a high-level of fault tolerance within one device because a secondary device provides it for you, in most instances.

• Any software configuration issue or software bug that results from an upgrade can be localized, assuming that only primary devices are upgraded first and the secondary devices are not.

• The primary and secondary devices do not need to be at the same location, which provides better redundancy in case of a catastrophic emergency, such as a fire.

• Layer 2 and Layer 3 protocols provide convergence. By picking the right protocol with the right feature, you can ensure quick convergence.

• A correct network design and implementation should enable you to use both the primary and secondary devices. For example, with STP, you can load-balance VLANs across uplink connections by having different root switches at the distribution layer by using PVST or MST.

Given these advantages, there are two main disadvantages to chassis redundancy: You need more data link layer connections, and because there are more devices, managing and troubleshooting this kind of network is much more difficult than with component redundancy.

It's important to point out that a well-designed redundant solution can include both types of redundancy. Remember that you need to determine what level of redundancy is required for the different components in your network.