VMan: Virtual Machine Management in Large-Scale Virtualized Computing Infrastructure


Background


Large-scale virtual computing infrastructures have become important platforms for many real world systems such as virtual computing lab (VCL), cloud computing, corporate data centers, and multi-tier web servers. Those infrastructures are expected to scale to tens of thousands of hosts and millions of virtual machines (VMs) in the near future. However, existing virtual infrastructures have inherent limitations pertaining to robustness, performance, and scalaliblity. One important reason attributed to the limitation is the absence of efficient distributed VM resource and performance management mechanisms. The goal of this project is to develop efficient and light-weight VM management techniques to greatly improve the scalability and resource-efficiency of the distributed virtual computing infrastructure. The proposed VM management system provides four essential mechanisms: 1) continuous monitoring of different VMs for detecting buggy or malicious guest VMs; 2) sharing-aware VM placement that allows multiple VMs to efficiently share host resources based on their load patterns; 3) adaptive VM ensemble expansion and contraction that allows an overloaded VM to clone itself into multiple instances for sharing workload or allows multiple duplicated VM instances to merge themselves when the workload drops; and 4) runtime VM migration that leverages live VM migration technology to continuously optimize hosted distributed application performance and balance resource utilization in the hosting infrastructure.
 

People


Faculty
Students

Collaborator

Publications


Related Projects

Code Release