Fault-tolerant can be defined as a computer system or unit devised so that, in case of any unexpected event of unit failure, a backup unit or method can instantaneously take over its role with zero loss of data or service. Fault tolerance is usually offered either along with software application, or is embedded in hardware configuration, or provided with some arrangement.
In the software application, the operating system provides an interface that enables the programmer to ""spot check"" the mission-critical data at pre-determined points within an operation. In the hardware application (for instance, with Stratus and its VOS operating system), the programmer does not need to be cognizant about the fault-tolerant capabilities of the system.
Generally speaking, fault tolerance is attained by duplexing each hardware component. Disks are emulated. Multiple processors are lock-stepped in unison and their outputs are equated to check the accuracy. When an incongruity ensues, the broken-down component is determined and replaced immediately making sure that the performance and functioning of the system doesn’t get affected at all.