E[X]-SJF[EsTLA]: A New Intelligent Rule Scheduling Approach in Active Database Systems
Abstract
Active database systems (ADBS) can react to the occurrence of predefined events automatically by defining a collection of active rules. One of the most important modules of ADBS is the rule scheduler which has considerable impact on performance and efficiency of ADBS. The Rule scheduler selects a rule to execute (evaluate) its action (condition) section in each time through the rules, which are ready for execution (evaluation). We have already evaluated and compared the existing rule scheduling approaches in a laboratory environment based on threetier architecture. Five evaluation criteria were recognized and defined formally for evaluation and comparison of rule scheduling approaches including: Average Response Time, Response Time Variance, Throughput, Time Overhead per Transaction and CPU Utilization. In this paper, we first design and implement the before mentioned laboratory environment again to cover and simulate the behavior of ADBS more exactly and completely, then propose a new approach to improve the rule scheduling process based on improvement of triggered rule scheduling using learning automaton. Then, we compare it with the most effective existing approach in the mentioned framework. Results of experiments show that the new method improves the rule scheduling.
Keywords
Estimation of Rule Execution Probability, Active Database Management Systems, Rule Scheduling, Learning Automaton