A fault-tolerant server provides a fully replicated hardware design that allows
uninterrupted service in the event of a component failure. Most components
have an active redundant partner. The recovery time or performance loss
caused by a component failure is close to zero, and all memory-based information
and disk contents are preserved. Fault-tolerant servers are expensive.
The problem with a fault-tolerant system is that the system itself has a single
point of failure (SPOF). Any software faults (such as application bugs or operating
system problems) can render the entire system unusable and result in service
outage.