QoS-driven resource management in real-time object based distributed computing systems

Quality-of-Service (QoS) driven resource management is a promising notion but has remained far from being established as a sound technology of a general form. The first step needed is to develop a practical and rigorous approach for specification of the QoS requirements associated with application functions. Such an approach must be accompanied by an approach for using such specifications as the main driver for cost-effective resource allocation. A promising concept, later named risk incursion function (RIF)-driven resource management, was created in 1994 and since then, slow efforts to convert the concept into a concrete technology have been under way. This research has been following another complimentary research on a promising middleware architecture that supports application software composed of distributed real-time objects and also manages execution resources for fault-tolerant system operations. In this paper, we discuss the current middleware structure into which mechanisms facilitating RIF-driven resource allocation can be incorporated with relative ease, as well as several practical patterns of RIF that have been established recently.