摘要：Many researchers around the world are working hard to improve cloud-based services focusing on server-side and communication aspects. This is especially true for services which demand high processing power and storage space on their server’s infrastructure, such as live video streaming on the web and video on demand. However, issues on client-side applications are often neglected or studied to a lower extent. In this context, this paper evaluates the occurrence of software aging in a web browser plug-in for video streaming. The case study is carried out by using an automated workload that simulates the user behavior accessing videos on YouTube. Time series were used to predict the resources utilization ahead of time, enabling the prevention of performance degradation and unexpected crashes. Finally, the prediction-based strategy is compared to the time-based strategy, showing the impact of both on system’s availability.
