Whether due to hardware or software faults, one fact cannot be disputed: network components fail. The only issues that can be debated are the frequency of the failure and the impact that each type of failure will have. HA is about mitigating the risks of network failures and bringing them within acceptable bounds, which are (or at least should be) dictated by your business strategy. Do you depend on your e-commerce Web site to be available 24 hours a day, 365 days a year? If so, your idea of acceptable network outages will be vastly different from someone whose business only relies on the network for sending and receiving occasional e-mails. Knowing your business strategy is the first step towards being able to decide which HA measures, if any, you should add.

The next step in justifying HA is to understand the trade-off between the cost and the improvement gained. Cost is not only measured in money, but also in time and complexity. A highly available network takes longer to implement, results in more maintenance work, takes longer to gain an understanding of, and due to increased complexity, raises the risks of human error. Making an informed decision is not as easy as you might think.

Depending on a businesses needs, HA can be anything from having a spare unit in the storage room to having a fully meshed, fully redundant network infrastructure with automatic failure detection and failover. Generally speaking, the term HA is...