A Systematic Approach to Deal with Noisy Neighbour in Cloud Infrastructure

Background/Objectives: One of the major challenges of the multitenant cloud model is performance unpredictability because of resource contention. The objective of this paper is to propose an approach to deal with noisy neighbours in a shared cloud infrastructure and reduce their affect on the other tenant applications. Methods/Statistical Analysis: Multiple tenant applications are deployed on cloud VMs which share the underlying system resources. Noisy neighbour applications take up more resources and leave other applications in turmoil state that leads to lack of predictability of the performance of other applications. The proposed system actively monitors the resource consumption of the tenant applications based on some identified parameters. Findings: Monitoring the resource consumption of the applications helps categorize the resource greedy applications as noisy neighbours. The other tenant applications which do not get their fair share of the resources are victim applications. Once the noisy neighbours have been identified, the next step is to deal with noisy neighbours by either migration of victim applications on a separate node on the cloud or borrowing resources from other nodes or implementing a quota system for resource allocation in cloud. Applications/Improvements: The pragmatic study of the behaviour of tenant applications based on their resource consumption pattern would help researchers to better focus on techniques to improve the Quality of Service of a cloud.