Variation-Aware Evaluation And Optimization For Cloud Workflow Resource Allocation

Due to the flexibility, convenience and low-cost maintenance of cloud computing, more and more workflow enterprises deploy their service to the cloud. However, influ-enced by uncertain factors such as manufacturing technique, variations (such as price variation and delivery time variation) exists in virtual machines. These variations can potentially influence execution performance, which makes it difficult for cloud service providers to stand by the time and cost constraint raised by customers. Variations will also cause the violence of Service Level Agreement and the decline of user experience.Although heuristics have been proposed to solve the problem of Cloud workflow resource allocation, it is hard to evaluate their performance under variations because of the lack of accurate modeling and evaluation methods.To address the above problems, a novel framework is proposed which can evaluate and optimize resource allocation strategies effectively and quantitatively based on statis-tical model checking and supervised learning. In this paper, the main contributions are as follows:1. Based on statistical model checking, an variation-aware modeling and evaluation approach is proposed for cloud workflow resource allocation.2. An automated negotiation approach is proposed to enable the tuning of parameters to improve the overall QoS, and thus to sign SLA.3. An efficient resource allocation optimization approach is proposed. Based on three mainstream supervised learning approach, the approach supports the quick opti-mization of overall resource allocation instances under customer requirements.The experimental results demonstrate that our framework can effectively support the evaluation and comparison of resource allocation strategies, the negotiation on Service Level Agreement (SLA) and efficient optimization on workflow resource allocation.