A Multi-Level Layered System Approach to On-Line Testing

Abstract

A significant body of work has been produced in the area of on-line testing of hardware. A large amount of this body of work originated in university research and has actually been adopted by industry to the point of permeating many of the current industrial products. Often, the techniques are well understood on a theoretical basis and constitute a legitimate science. On the other hand, it has been more difficult to establish the same level of scientific rigor in higher levels of on-line testing (unit, system, and software). Paradoxically, however, it is essential to take care of errors and failures at all level of the system hierarchy, inclusive of the application software level. This is particularly true for systems, which must guarantee continuous service delivery in spite of errors and malfunctions. As a result, the industry has implemented, for higher system levels, empirical methods, which are often not very well documented. This situation is less prevalent in the area of general purpose computing, but it is quite frequent in domain oriented real-time systems, such as telecommunication equipment. This paper discusses a few examples of on-line testing empirical techniques used in the telecommunication industry, and presents a few attempts at more rigorous approaches. It also outlines important directions for additional research.