On Quality-of-Service Prediction And Optimal Scheduling Of Cloud Computing Systems In Unreliability Environment

Cloud computing is a hot topic in recent years.It arises as a new IT service model,which can provide people credible,cheap computing resource.In cloud environment,different type of resource are virtualized to conduct unified management and presented as various kinds of resource pool.In this model,the cloud Datacenter can deploy tasks submitted automatically,cloud users can use the computing resource without purchasing extra physical devices and withdraw from maintaining.To build such a platform,there must be an excellent algorithm to solve the problem that how to schedule the resource for tasks.The cloud computing paradigm enables elastic resources to be scaled at run time satisfy customers' demand.Cloud computing provisions on-demand service to users based on a pay-as-you-go manner.This novel paradigm enables cloud users or tenant users to afford computational resources in the form of virtual machines(VMs)as utilities,just like electricity,instead of paying for and building computing infrastructures by their own.Performance usually specified through Service-Level-Agreement(SLA)performance commitment,of clouds is one of key research challenges and draws great research interests.Actually,resource scheduling strategy is to build the mapping relationships between resource and tasks.There are two levels in the scheduling model: one is to schedule virtual machines for cloud tasks and another is to find suitable hosts to load virtual machines.A well performed strategy mainly decides the performance of the platform and it should reach some goals,such as user QoS,minimizing execution time,load balance,and economical efficiency.Quality-of-Service of cloud computing systems is receiving increasing attention from both the industry and academy.Most existing studies in this direction concern the formalization and property-verification issues.However,quantitative and nonfunctional QoS models and methods are still limited and preliminary in many ways.Comprehensive QoS models capable of describing fine-grained control-flow evolutions and predicting run-time QoS are still in high need.Motivated by the above-mentioned observations,this project presents a model-driven framework to quantitative study of QoS of Infrastructure-as-a-Service clouds.Detailed efforts to be carried out include:(1)developing stochastic models for fine-grained description of static architecture of cloud computing systems and dynamic control-flow evolution of cloud tasks;(2)modeling run-time cloud resource management and scheduling strategies(e.g.,dynamic VM migration,speed scaling,VM consolidation,transactional rollback)and deriving closed-form/product-form solutions of multiple QoS metrics;(3)employing time-series-based approaches to model the historical fluctuating QoS data and predicting future trend;(4)employing bio-inspired methods and algorithms to dynamically schedule workflow-based service processes on cloud infrastructures for performance optimization and cost reduction purposed.The study also includes case studies based on real-world cloud services,where experimental QoS results are used to validate the correctness and accuracy of proposed methods.