Ph.D., Colorado State University, December 2008
Co-Major Professors: H. J. Siegel and Anthony A. Maciejewski
This research investigates the problem of robust resource allocation for distributed computing systems operating under imposed Quality of Service (QoS) constraints. Often, such systems are expected to function in a physical environment replete with uncertainty, which causes the amount of processing required over time to uctuate substantially. In the first two studies, we show how an effective resource allocation can be achieved in the heterogeneous shipboard distributed computing system and IBM cluster based imaging system. The general form for stochastic robustness metric is then presented based on a mathematical model where the relationship between uncertainty in system parameters and its impact on system performance are described stochastically. The utility of the established metric is exploited in the design of optimization techniques based on greedy and iterative approaches that address the problem of resource allocation in a large class of distributed systems operating on periodically updated data sets. One of the major reasons for possible QoS violations in distributed systems is a loss of resources, frequently caused by abnormal operating conditions. One aspect that makes a resource allocation problem extremely challenging in such systems is a random nature of resource failures and recoveries. The last study presented in this work describes a solution method that was developed for this case based on the concepts of the Derman-Lieberman- Ross theorem. The experimental results indicate a significant potential of this approach to generate robust resource allocations in unstable distributed systems.