TOWARD INTEGRATION OF MAJORDESIGN TECHNIQUES FOR REAL-TIMEFAULT-TOLERANT COMPUTER SYSTEMS

摘要

Existing major design techniques for real-time fault-tolerant (RTFT) computer systems remainrnunsatisfactorily integrated although they are complementary in nature. There are severalrnchallenging issues that must be resolved in order to pave ways to achieve cost-effective integrationrnof those design techniques. This paper starts with the introduction of a scheme for classificationrnof various fault tolerance schemes according to the benefits that they bring. Thereafter, a reviewrnof important design techniques that have broad applicability in designing RTFT computerrnsystems follows. Included among the reviewed techniques are the techniques for designingrnfault-tolerant processing nodes capable of real-time forward recovery from hardware and/orrnsoftware faults and the techniques for real-time network surveillance and reconfiguration. Issuesrnin integrating these sets of techniques are then discussed. Another set of design techniquesrnwhich will play increasingly important roles in future engineering of RTFT computer systemsrnare those for object-oriented (OO) structuring. Issues to be resolved in and possible directionsrnfor integrating the important fault tolerance techniques with the OO structuring techniques arerndiscussed next.