Design And Implementation Of Real-time Distributed System Fault-tolerant Scheduling Algorithm

Task scheduling in Real-time Distributed Computer System (RTDCS) is anissue that is full of challenge, also a hot research subject nowadays. Because taskscheduling problem is a typical NP problem and a critical aspect which directlyimpacts the system performance, to study task scheduling scheme in RTDCS is ofgreat significance. With the support of the key national pre-research project for the10th five-year plan, this paper has deeply studied scheduling theory andimplementations of RTDCS, which arms to improve resource utilization bycombining with processor scheduling algorithms and a lot of real-time period tasksin real life.Firstly, this paper describes the module and characteristics of concerningsystems with Real-Time Distributed Fault-Tolerant System (RTDFTS). Afteranalyzing scheduling theory, we point out that the redundancy of RTDCS makefault-tolerant possible. Further more, we give a certification for schedulablity oftasks and its criteria for performance evaluation.Secondly, according to fault-tolerant scheduling theory of distributed systems,we present two different real-time distributed fault-tolerant scheduling algorithms:PPB-MFT-EDF and APB-MFT-EDF. The first one is passive backup, while last oneis active backup. In this paper, a certification for task schedulable is given and someimportant character about the two algorithms is present. The certification andsimulation results have prove that the performance of APB-MFT-EDF algorithm isbetter than PPB-MFT-EDF.Further more, we put forward a mixed tasks scheduling algorithm that canprocess periodic and aperiodic tasks. By converting aperiodic tasks into periodictasks and utilizing the free resource effectively, this algorithm is quite simple andavailable for different types of task. Combining with processor fault-tolerantscheduling, a new algorithm that implements a multi type fault-tolerant function ispresent and the resource utilization is improved.In the end of the paper, based on a kind of architecture of RTDFTS and threefault-tolerant methods, we present a structure chart of dynamic task assigning and...