Dependable Computing

2Classical Dependability Techniques and Modern Computing Systems: Where and How Do They Meet?

2.1 Illustrative Case Studies of Design for Dependability

In this chapter, we attempt to engage the interest of the reader by beginning with a brief discussion of examples of real systems to illustrate the deployment of traditional techniques in legacy systems. The examples show how fault tolerance needs to be considered as part of the design requirements, starting in the initial stages of a system development cycle. The first example, IBM System/360 [1, 2] (which ran on the proprietary IBM MVS operating system) introduced many dependability techniques and was a commercial success. The design of the system presents an instructive example of the use of fault tolerance techniques in a traditional architecture. The second example, the Tandem Integrity system [3, 4] was the first commercial, loosely synchronized, TMR (Triple Modular Redundancy), fault‐tolerant system that ran the UNIX operating system, which made it possible to support many standard applications on a high‐availability platform. The Integrity system succeeded in bringing together many methods that were part of sophisticated fault‐tolerant systems to establish something that approximated a commodity environment. The third example is Blue Waters, a petaflop HPC system capable of delivering approximately 13.1 petaflops (at peak) for a range of real‐world scientific and engineering applications.

2.1.1 IBM System S/360

During ...