PROJECT TITLE :

Optimizing for Tail Sojourn Times of Cloud Clusters - 2018

ABSTRACT:

A standard pitfall when hosting applications in these days's cloud environments is that virtual servers often experience varying execution speeds because of the interference from co-located virtual servers degrading the tail sojourn times specified in service level agreements. Motivated by the importance of tail sojourn times for cloud clusters, we have a tendency to develop a model of N parallel virtual server queues, every of that processes jobs in a very processor sharing fashion below varying execution speeds ruled by Markov-modulated processes. We derive the tail distribution of the workloads for each server and also the approximation for the tail sojourn times primarily based on massive deviation analysis. Furthermore, we tend to optimize the cluster sizes that fulfill the wants of target tail sojourn times. Extensive simulation experiments show very good matches to the derived analysis in a selection of situations, i.e., massive numbers of servers experiencing a high range of various execution speeds, below numerous traffic intensities, workload variations and cluster sizes. Finally, we tend to apply our proposed analysis to estimating the tail sojourn times of a Wikipedia system hosted in an exceedingly personal cloud, and also the testbed results strongly make sure the applicability and accuracy of our analysis.