Tool-driven strategies for resource provisioning of single-tier web applications in clouds

The rapid growth in size and complexity of web applications makes it difficult to predict their resource demands. Consequently, large web server clusters are needed to guarantee Quality of Service (QoS) requirements and this leads to a waste of resources. The on-demand and elastic nature of cloud computing has opened a new solution for this problem. A web application can be deployed in clouds in such a way that its resource demand can be scaled dynamically to adapt to workloads from user requests, where the cost benefit can only be obtained with efficient resource provisioning strategies. In this paper, we introduce an evaluation tool for the effectiveness of resource provisioning strategies, based on queuing theory. Using this tool, we propose a prediction-based resource provisioning strategy which provisions resources more efficiently than common strategies.