Techniques for fault detection, reconfiguration, and computation recovery in distributed computing systems have been slowly refined over at least two decades. Both the types and the fault rates of replaceable modules have continuously changed. For example, the logical complexity of the replaceable modules has steadily grown while their physical sizes may have remained unchanged or shrunken. Interests of industry have shifted accordingly.

In the work-group server and data server field, basic techniques for maintenance of data integrity were established long time ago and have been slowly refined. Also, use of RAID (redundant array of inexpensive disks) controllers has increased and down-times of network switches and routers have been greatly reduced over the past decade.

In real-time (RT) computing (RTC) applications, the requirements on fault tolerance have been significantly more severe than those in non-RT data processing environments. Demands for abilities for analysis and validation of the impacts of fault tolerance mechanisms on systems, i.e., overhead during fault-free operation, fault detection latency, and worst-case increases in system response time, have always been there but the technology has been slowly advancing and remains far from the mature state. This is mainly because the technology for designing basic, non-fault-tolerant, RTC application systems yielding analyzable and yet tight response time bounds has not matured yet. Therefore, even though the fault-tolerant computing (FTC) approaches which will not aggravate the response time bound analyzability to any significant extent have been formulated, opportunities for inserting them during RTC application system design have been rare.