A novel pure adaptive failure detector for dynamic distributed system

摘要

Performance level of dynamic distributed computing systems are found to suffer due to failure of multiple nodes or processes during its operation. Accuracy and speed are the two major QoS requirements governing their timely detection and hence to sustain the performance level of the distributed system. Several adaptive failure detectors based on grey theory have been proposed by earlier investigators. They suffer from their inability to address the cited requirements due to the issues related to fluctuating dynamic behavior of distributed systems. The trend and potency associated with the dynamism of modern day distributed system varies as a function of time and factors related to the same appears to have not been resolved so far. The present work proposes a failure detection approach for highly dynamic distributed computing systems wherein issues related to potency and trend of dynamism are addressed by employing an inverse recursive approach. The approach complied in the form of algorithms serve as a precompiled code in executing the detection task of multiple node failures in a dynamic distributed system with enhanced accuracy and speed and the experimental results validates the same.