Challenging the Robustness of Self-Managing Computing Systems for QoS Controller Design

Abstract

A widespread interest in lowering the requirement for human involvement by enabling systems to operate autonomously has been sparked by the high expense of running massive computing installations. The extent to which control theory can offer an analytical and architectural basis for developing self-managing systems is examined in the present research. For the QoS management of such systems, these methods often employ fixed, adaptive, or single model based control techniques. However, it is challenging to create a single model or controller that would provide the required QoS performance over all of these systems' operational zones due to the variable system dynamics and unforeseen environmental changes. The goal of a novel approach known as self-managing computer systems is to incorporate into the systems the means by which they can automatically modify configuration parameters to ensure that the system's Quality of Service requirements are continuously satisfied. In this research, we assess the resilience of such approaches under high variable workloads in terms of request service times and inter-arrival times. This research also makes a contribution by evaluating how workload forecasting techniques are used in QoS controller design.