A Run-Time Adaptable Persistency Service using the SMART Framework

Run-time adaptation is becoming an important feature to increase robustness, performance, and improve resource utilization. A system using a persistency data service can benefit from this feature by avoiding constraint violations. This aids in averting system failures and consequent data loss. The application and assessment of a run-time adaptable approach, the SMART (State Model Adaptive Run-Time) Framework, to a data persistency service is the subject of this paper. SMART is a component-based approach with adaptable features founded on the mathematics of control theory and fuzzy logic. The results of the study indicate that the use of SMART to the persistency data service avoided constraint violations in 95% of the cases over an execution period of more than eight hours. These results are encouraging and provide strong evidence of the advantages of using the SMART Framework.