Business drives high-availability requirements

High availability (HA) requirements should be driven by business and not implemented for the sake of technology.

Seeking to use a new high availability technology, supporting a clustered file system, etc. may be interesting and intellectually satisfying. The most important consideration is that these technologies advances the business goals without making the system overly complicated or expensive.

In some cases, the business may be able to tolerate a certain amount of outage and a simple backup and restore may suffice. Of course, there are others where an hour of downtime is very expensive and as a result require that every component from power to the database be made as resilient as possible through redundancy and automated failover. In addition, they may mandate geographically dispersed disaster recovery capabilities.

High availability designs cannot be performed in isolation. As in most worthy engineering endeavors, high availability requirements must be balanced against other architectural choices including acquisition cost, maintenance costs, scalability, maintainability, ease of use, impact to business, and so forth.