With the proliferation of hosting and co-location facilities both locally and on the mainland, many businesses and government organizations are adapting these sites to support replication of their critical computer systems. That is, folks keep a copy of their computer system at a commercial location, as a standby in case they experience a failure. While redundancy is a great idea, there are options to consider when implementing such a solution.

Traditionally there are three options when using a standby system. Typically referred to as “hot,” “warm” and “cold,” these temperature ratings correlate to the readiness of the replicated system.

A hot standby system runs at all times in conjunction with and parallel to the main or production system. It is virtually identical to the production system in terms of hardware and software configuration and operation. The hot standby contains data that is reasonably current, ideally no more than a few seconds old, although it’s not uncommon to see data a few minutes old. Switch-over from the production to the hot standby also is done very quickly, again, ideally in less than a few seconds but sometimes as long as a few minutes.

Like a hot standby, a warm standby is operational at all times. A warm standby is not necessarily identical to the production system, but is capable of running all necessary functions. Such a system typically takes longer to bring online than a hot standby, sometimes a few hours, and usually requiring a data update.

A cold standby is merely hardware that is configured to be able to run the production system software, but is not running. A cold standby typically require data and sometimes application updates to be made before becoming usable. A cold standby is usually brought online less than 24 hours after the main system goes down, but sometimes can take as long as 72 hours.

As technology has evolved, so have options of system replication. For example, some cold standby processes employ a strategy to copy data in real time, similar to a hot standby. This facilitates a much faster system restoration process than the traditional cold standby.

In terms of cost and complexity, a hot standby is the most difficult to engineer and the most expensive. A cold standby is the cheapest and easiest to deal with, while a warm standby falls someplace in the middle.

Regardless of the solution you choose, it is imperative that it be tested periodically. Referred to as a fail-over or rollover test, this is the only way to tell for sure that your standby system is working properly. Some organizations may even choose to fully exercise their standby system by running on it for 30 days or more.

John Agsalud is an IT expert with more than 25 years of information technology experience. Reach him at johnagsalud@yahoo.com.