Improved AED Scheduling Algorithm for Real-Time System

The design and implementation of real-time database presents many new challenging problems. Compared with conventional database, real-time database have distinct features: they must maintain the coherent data while satisfy the timing constraints associated with transaction. In addition, real-time database must adapt to changes in operating environment and guarantee the completion of critical transaction with favoring changes in the system. With evolution of Earliest Deadline First (EDF) in 1973 by LIU and LAYLAND, laid the path for development of RTDB, it is very inefficient in overloaded workload conditions. Adaptive Earliest Deadline (AED) improves the performance which uses feedback control mechanism to detect overloaded condition. There prevails the risk of losing transaction with extremely high value may cause severe losses to system. A extension of AED called Hierarchical Earliest Deadline (HED) provide solution by establishing the value based bucket hierarchy thus ensuring the completion of high value transaction. Where value assigned reflects the return expected to receive if the transaction commits before its deadline. In this paper we present the comparison of EDF, AED under different workload conditions with calculated examples. We are also proposing a method which will focus on using analysis of arrival-time of transaction, seek-time and transaction size to determine best scheduling algorithm for the current workload, switching and tuning algorithm as necessary to improve performance.