Using Control Theory for Load Shedding in Data Stream Management

Database performance can be greatly affected by environmental and internal dynamics such as workloads and system configurations. Existing strategies to maintain performance under such dynamics are often found to have poor robustness. To remedy this problem, we propose a systematic solution that takes advantages of formal feedback control techniques. In this demo, we show how the control-based solution derived from a dynamic DSMS model can be utilized to guide load shedding with the target of maintaining data processing delays.