Model Driven Engineering for Quality of Service Management: A Research Note on the Case of Real-Time Database Management Systems

Qos Management In Rtdbms

Real-time transactions manipulating real-time data can be update or user transactions. Update transactions have to periodically update real-time data which have validity periods beyond which they become not fresh. The consequences of accessing data outside their validity periods depend on particular requirements of the application and data semantics. User transactions may access real-time data and update non real-time data. They must be processed within their deadlines and use fresh data (Ramamritham, Son & Dipippo, 2004). Moreover, they have unpredictable arrival and execution times. In this context, the RTDBMS may face unpredictable overload periods over time, during which its performances may be downgraded and many transactions may miss their deadlines.

To address this problem, many QoS Management Solutions have been proposed and most of them have used the Feedback Control Scheduling Architecture FCSA (Lu, Stankovic, Tao & Son, 2002). The first sub-section explains how to specify the QoS in RTDBMS and the operating principle of the FCSA architecture. The second sub-section presents some QoS solutions for RTDBMS.