When talking about a cluster or a system with load balancing, High Availability (HA) is also often mentioned. A HA system is a system that, as a whole, is available without interruption at almost all times. The almost here is the fresh wind of reality kicking in. In practice, it can be enormously costly—if at all possible—to construct a system with 100 percent availability. It is, therefore, common to talk about the number of nines the system is available to. Four nines means 99.99 percent, five means 99.999, and so on. The number of nines is an often-recurring attribute in Service Level Agreements (SLA).

In a HA system, the resilience to disturbances is imperative. Common solutions are to have more nodes alive than necessary or to have the possibility of quickly starting up new ones. A virtualized environment becomes increasingly utilized for these purposes and for quickly scaling up or down according to how the load changes over time (per hour, day, or month).