ABSTRACT
Fault tolerance is the ability of a system to continue correct operation after the occurrence of hardware or software failures or operator errors. The intended system application is what determines the system reliability requirement. Since computers are used in a vast variety of applications, reliability requirements differ tremendously. For very low-cost systems, such as digital watches, calculators, games, or cell phones, there are minimal requirements: the products must work initially and should continue to operate for a reasonable time after purchase. Failures of these systems are easily discovered by the user. Any repair may be uneconomical. At the opposite extreme are systems in which errors can cause loss of human life. Examples are nuclear power plants and active control systems for civilian aircraft. The reliability requirement for the computer system on an aircraft is specified to be a probability of error less than 10−9
per hour [hissa.nist.gov/chissa/SEI Framework/framework 7.html]. More typical reliability requirements are those associated with commercial computer installations.
For such systems, the emphasis is on designing system features to permit rapid and easy recovery from failures. Major factors influencing this design philosophy are the reduced cost of commercial off-theshelf (COTS) hardware and software components, the increasing cost and difficulty of obtaining skilled maintenance personnel, and applications of computers in banking, on-line reservations, networking, and also in harsh environments such as automobiles, industrial environments with noise sources, nuclear power plants, medical facilities, and space applications. For applications such as banking, on-line reservations, or e-commerce, the economic impact of computer system outages is significant. For applications such as space missions or satellites, computer failures can have a huge economic impact and cause missed opportunities
to record valuable data that may be available for only a short period of time. Computer failures in industrial environments, automobiles, and nuclear power plants can cause serious health hazards or loss of human life.
