SLA-aware resource management for application service providers in the cloud

In the today Internet of Services, one of the challenges of Application Service Providers (ASPs) is to fulfill the QoS requirements stated in the Service Level Agreements (SLAs) established with different consumers and to minimize the investment and management costs. Cloud computing is the promising solution for ASPs that increasingly demand for an elastic infrastructure. In this paper, we formulate the ASP resource management as an optimization problem and propose both reactive and proactive heuristic policies that approximate the optimal solution. The proposed policies leverage on information about the system performance history and can be applied at runtime because of their reduced computational time. Our experimental results show that some heuristics based on prediction approximate the exact knowledge of the workload.