4.5. FAULT TOLERANCE
A system is said to fail when it does not meet its specification. In some cases, such as a supermarket's distributed ordering system, a failure may result in some store running out of canned beans. In other cases, such in a distributed air traffic control system, a failure may be catastrophic. As computers and distributed systems become widely used in safety-critical missions, the need to prevent failures becomes correspondingly greater. In this section we will examine some issues concerning system failures and how they can be avoided. Additional introductory material can be found in (Cristian, 1991; and Nelson, 1990). Gantenbein (1992) has compiled a bibliography on the subject.