Scheduling Partially Ordered Events in a Randomised Framework: Empirical Results and Implications for Automatic Configuration Management

• Authors: Frode Eika Sandnes
• Venue: LISA '01 Proceedings of the 15th USENIX conference on System administration
• Year: 2001

Automatic configuration management involves maintaining a set of shared and distributed resources in such a way that they serve a community of users fairly, promptly and reliably. In this context, this paper discusses experiments that measure the effect of adding randomized scheduling of partially ordered events to configuration management tools. Three characteristics of randomized scheduling are investigated: efficiency, robustness and security. A configuration management process is efficient if it minimizes the use of resources. It is robust if it is not vulnerable to malicious acts or inadvertent human errors. It is secure if its management model is hidden from observers. Several experiments suggest that randomized scheduling of partially ordered events has advantages over commonly used deterministic strategies, on average producing more efficient schedules. Further, randomized scheduling greatly degrades the accuracy of observer predictions of future behavior. In addition, randomized scheduling obscures the management model such that an observer will have to make a large number of observations in order to obtain the complete management model. The results of the study support the use of randomization in automatic configuration management tools.