Assessment of Aging and Rejuvenation for Resiliency in Heterogeneous Network Clusters

The problem of evaluating the application of software rejuvenation is challenging, especially when load balancing comes into play to overcome resiliency threats. In this paper, we focus on the evaluation of the software aging threat due to memory leaks in the underlying open-source infrastructure. Our goal is to evaluate the effectiveness of several software rejuvenation strategies to achieve resiliency. To this end, we have built a heterogeneous environment composed of 15 large virtual machines and 15 extra-large virtual machines. We have used an open-source benchmark, and we have measured the benchmark response time on the large heterogeneous network. As a result, we obtained a software rejuvenation threshold that was used to evaluate the extent of software aging in each virtual machine.We have created an automated testing and analysis infrastructure that was used to evaluate several strategies for achieving resiliency for large infrastructure subject to software aging. We present the obtained results of the application of software rejuvenation to maintain the resiliency of a large heterogeneous network of virtual machines.