On the Reflective Nature of the Spring Kernel

Abstract

The Spring Kernel is a research oriented kernel designed to form the basis of a flexible, real-time operating system for complex, next generation, real-time applications. The Spring Kernel is being implemented in stages on a network of 68020 and 68030 based multiprocessors called SpringNet. Version 1 of the Kernel is now operational. While much has already been written on the Spring Kernel [10,11,12,4], the purpose of this invited paper is to combine ideas found in separate papers and presents them with a different emphasis. In particular, in Section 2.1 we categorize real-time systems to clearly indicate the difficulty in building such systems. Since most real-time systems aspire to being predictable, we discuss exactly what predictability means for such systems. We show that it is important to fully understand the implications of predictability and to not over estimate its value. In Section 2.2 we discuss how the notion of predictability can likely be achieved by an integrated approach to designing and building next generation hard real-time systems. Much of Section 2 is taken from [16]. In Section 3 we review the main ideas found in the Spring Kernel. Much of the material found in Section 3 is from [12]. In Section 4 we discuss the reflective nature of the Kernel providing the emphasis that has not appeared in previous papers. By reflection is meant the ability of the Kernel to maintain and act on information concerning the application, the environment, and the Kernel itself. This includes identifying what information is to be used, how to monitor this information, and how to dynamically adapt the system. Section 5 makes summary remarks.

This work was supported by ONR under contract NOOO14–85-K-0389 and NSF under grant DCR-8500332.