Experimental research in dependable computing at Carnegie Mellon University - From Faults to Manifestations

In 1945, the Carnegie Plan for higher education was evolved. The basic philosophy of the plan was "learning by doing". The strong emphasis on experimental research at Carnegie Mellon University (CMU) is one example of the Carnegie plan in operation. In particular, research in reliable computing at CMU has spanned five decades of researchers and students. In the early 1960's, the Westinghouse Corporation in Pittsburgh had an active research program in the use of redundancy to enhance system reliability; William Mann, who had been associated with CMU, was one of the researchers involved in this effort. In 1962, a symposium on redundancy techniques was held in Washington, D.C., and led to the first comprehensive book [57], co-authored by Mann, on the topic of redundancy and reliability. CMU's Professor William H. Pierce wrote a paper on adaptive voting [40] that formed a part of this book. Pierce also published one of the first textbooks on redundancy [41]. During the next four decades, a large number of experimental hardware and software systems were designed, implemented, and made operational, at CMU. These systems covered a range of computer architectures, from uniprocessors, to multiprocessors, to networked systems. Each system represented a unique opportunity to include, and to quantify the results of incorporating, reliability features in the design of the system. This paper surveys the monitoring, measurement, and evaluation of those systems. A common theme has been to understand the natural occurrence of faults, to develop mathematical models for prediction, and to raise the level of abstraction of fault-models in order to monitor and design dependability mechanisms more easily. Figure 1 illustrates the breadth, diversity, depth and chronological progress of the dependability research at CMU, over the past few decades up until the present day.