QoS-aware multi-cloud brokering for NGN services: Tangible benefits of elastic resource allocation mechanisms

摘要

Elastic mechanisms for rapidly scaling cloud resources according to dynamic workloads are essential characteristics of cloud computing. The dynamic selection of current optimal cloud services is one of the main objectives of cloud brokering mechanisms. Combined with capabilities for elastic cloud resource allocation, multi-cloud brokering mechanisms allow for dynamic selection and efficient utilization of cloud resources. The achieved efficiency of resource consumption under dynamically changing workloads and different QoS constraints is a key performance indicator of such systems. This paper evaluates the effects of multiple factors impacting the performance of such systems in a large-scale multi-cloud environment for research and experimentation of Future Internet technologies. It defines metrics for assessing the performance of multi-cloud brokering systems, against which a cloud brokering prototype for cloud-based NGN media streaming services is evaluated. By evaluating the impact of application- and cloud resource-specific effects, as well as the specific characteristics of realistic NGN workloads, a realistic estimation of potential cost-/resource savings and achievable QoS levels is provided.