Predicting Software Abnormal State by using Classification Algorithm

Software aging, also called smooth degradation or chronics, has been observed in a long running software application, accompanied by performance degradation, hang/crash failures or both. The key for software aging problem is how to fast and accurately detect software aging occurrence, which is a hard work due to the long delay before aging appearance. In this paper, two problems about software aging prediction are solved, which are how to accurately find proper running software system variables to represent system state and how to predict software aging state in a running software system with a minor error rate. Firstly, the authors use proposed stepwise forward selection algorithm and stepwise backward selection algorithm to find a proper subset of variables set. Secondly, a classification algorithm is used to model software aging process. Lastly, t-test with k-fold cross validation is used to compare performance of two classification algorithms. In the experiments, the authors find that their proposed method is an efficient way to forecast software aging problems in advance.

Failure Detectors of Strong S and Perfect P Classes for Time Synchronous Hierarchical Distributed Systems

Present failure detection algorithms for distributed systems are designed to work in asynchronous or partially synchronous environments on mesh (all-to-all) connected systems and maintain status of every other process. Several real-time systems are hierarchically connected and require working in strict synchronous environments. Use of existing failure detectors for such systems would generate excess computation and communication overhead. The chapter describes two suspicion-based failure detectors of Strong S and Perfect P classes for hierarchical distributed systems working in time synchronous environments. The algorithm of Strong S class is capable of detecting permanent crash failures, omission failures, link failures, and timing failures. Strong completeness and weak accuracy properties of the algorithm are evaluated. The failure detector of Perfect P class is capable of detecting crash failures, crash-recovery failures, omission failures, link failures, and timing failures. Strong completeness and strong accuracy properties of the failure detector are evaluated.

Performance Optimization of Persistent Memory Systems Through Phase-Based Transactional Memory

The emerging persistent memory technologies (PM) are aimed to eliminate the gap between main memory and storage. Nevertheless, its adoption requires measures to guarantee consistency, since crash failures might render the program in an unrecoverable state. In this context, the usage of durable transactions is one of the main investigated approaches to ease the adoption of PM. However, today's implementations are based exclusively on software (SW) or hardware (HW), which might degrade system performance. This paper presents NV-PhTM, a transactional system for PM that delivers the best out of both HW and SW transactions by dynamically changing the execution according to the application's characteristics.

Failure Detectors of Strong S and Perfect P Classes for Time Synchronous Hierarchical Distributed Systems

Present failure detection algorithms for distributed systems are designed to work in asynchronous or partially synchronous environments on mesh (all-to-all) connected systems and maintain status of every other process. Several real-time systems are hierarchically connected and require working in strict synchronous environments. Use of existing failure detectors for such systems would generate excess computation and communication overhead. The chapter describes two suspicion-based failure detectors of Strong S and Perfect P classes for hierarchical distributed systems working in time synchronous environments. The algorithm of Strong S class is capable of detecting permanent crash failures, omission failures, link failures, and timing failures. Strong completeness and weak accuracy properties of the algorithm are evaluated. The failure detector of Perfect P class is capable of detecting crash failures, crash-recovery failures, omission failures, link failures, and timing failures. Strong completeness and strong accuracy properties of the failure detector are evaluated.