Coordinated QoS-Driven Management of Cloud Computing and Storage Resources

Project Details

Project Lead
Ming Zhao 
Project Manager
Ming Zhao 
Institution
Florida International University, School of Computing and Information Sciences  
Discipline
Computer Science (401) 

Abstract

To ensure that cloud users are comfortable with running their critical applications on shared infrastructures and that cloud providers can support economical performance-based service-level agreements, there is an increasingly urgent need for virtualized systems to deliver performance guarantees. And, as virtualization and cloud computing become pervasive, it is also important that these topics are taught in a systematic manner; especially to minority students served by this PI's university. In order to address these challenges, this NSF CAREER project is creating a coordinated resource management framework that optimizes the allocations of cloud computing and storage resources according to application-desired Quality of Service (QoS).

Intellectual Merit

Specifically, this project is accomplishing its objectives through the following three research and education components: 1) A QoS-driven virtual machine resource management framework that can coordinate the allocations of various computing and storage resources and optimize them according to the virtualized applications; 2) QoS-driven distributed virtual machine storage management that allows the allocation of shared cloud storage resources, including the emerging solid-state-drive-based virtual machine storage and caching, according to application QoS needs; 3) Systematic education on virtualization and cloud computing that harnesses the research outcomes to provide training in virtualization and cloud computing, including new education activities for graduate, undergraduate, and K-12 students, as well as a new virtual-machine-based online education system to facilitate these activities.

Broader Impacts

This project's research outcomes will enable virtualized systems to support performance guarantees for modern applications with dynamic and complex behaviors. As a result, a broader range of applications with different QoS requirements will benefit from cloud computing, and cloud services will be able to offer their users more economical QoS-based charging models instead of the currently used resource-capacity-based models. This project's education outcomes will enable systematic education on virtualization and cloud computing from K-12 to undergraduate and graduate classrooms and prepare a pipeline of students who are equipped with the necessary knowledge and skills in these emerging technologies and prepared to contribute in the coming cloud computing era.

Scale of Use

100 to 200 VMs for one to two months

Results