Handling Overload and Data-Relaxation Control in Distributed Real-Time Database Systems

Current applications are distributed in nature and manipulate time-critical databases with firm-deadline transactions. To guarantee performance and availability of such applications fault-tolerant mechanisms need to be developed. In the context of time critical applications, this also leads to providing means to ensure that as many as possible real-time transactions satisfy their deadline in the presence of overload. In this paper, we propose a distributed protocol, which provides mechanisms to handle overload situations while tolerating bounded data inconsistencies in the context of distributed real-time database systems.