Journals Proceedings

Abstract

Innovations in electronics and hardware have led to the scenario of multiple softwares running on the same hardware resulting in multiuser, multitasking and virtualized environments. The reliability of such high performance computing systems depends both on hardware and software. The software systems, during operation accumulate errors or garbage leading to software aging which may lead to system failure and hazardous consequences. Software aging needs to be dealt with specialized techniques. To deal with software aging, a technique called software rejuvenation exists that reboots or re-initiates the software to avoid fault or failure. The detection of software aging is one of the important steps before rejuvenating the system as the rejuvenation process may lead to downtime which has a business impact. Hence, software aging prediction is gaining importance and is one among the trends in the recent research area in the field of software aging and rejuvenation. In this paper, it is explored how the machine learning technology can be used to predict software aging. The predictive power of the built in algorithms of machine learning tools can be used for software aging forecasting. The software aging model has been built using machine learning tools. The developed software aging detection model can be used to schedule the rejuvenation to prevent crash or performance degradation of software systems.