QoSCloud

QoS-driven Virtualized Computing Resource Management

System virtualization is an increasingly powerful technology that enables the emerging computing paradigms such as public and private cloud systems. It allows applications to be conveniently deployed along with their required execution environments through virtual machine (VMs), and supports them to flexibly share the underlying physical resources with strong isolation. However, there exists an increasingly urgent need for virtualized systems to deliver strong Quality of Service (QoS) guarantees to their hosted applications. Currently such systems can meet only coarse-grained and relaxed performance requirements, and their management considers only limited facets of an application's multi-type resource usage. As a result, examples such as cloud systems cannot support QoS-based Service Level Agreements (SLA) with their hosted applications. The continued existence of the lack of strong QoS guarantees from virtualized systems presents a critical hurdle to their further adoption by applications and their support of more economical QoS-based SLAs. The objective of this project is to create a QoS-driven multi-type resource management system to support strong QoS guarantees for applications hosted on virtualized computing systems.

Resource management in virtualized systems remains a key challenge because of their intrinsically dynamic and complex nature, where the applications have dynamically changing workloads and virtual machines (VMs) compete for the shared resources in a convolved manner. To address this challenge, this project proposes a new resource management approach that can effectively capture the nonlinear behaviors in VM resource usages through fuzzy modeling and quickly adapt to the changes in the virtualized system through predictive control. The resulting fuzzy-model-predictive-control (FMPC) approach is capable of optimizing the VM-to-resource allocations according to high-level service differentiation or revenue maximization objectives.

Existing resource management solutions in datacenters and cloud systems typically treat VMs as black boxes when making resource allocation decisions. This project advocates the cooperation between VM host- and guest-layer schedulers for optimizing the resource management and application performance. It presents an approach to such cross-layer optimization upon fuzzy-modeling-based resource management. This approach exploits guest-layer application knowledge to capture workload characteristics and improve VM modeling, and enables the host-layer scheduler to feedback resource allocation decision and adapt guest-layer application configuration.

Publications

• Lixi Wang, Jing Xu, Ming Zhao, "Modeling VM Performance Interference with Fuzzy MIMO Model," 7th International Workshop on Feedback Computing (FeedbackComputing, co-held with ICAC2012), September 2012.
• Lixi Wang, Jing Xu, Ming Zhao, "Application-aware Cross-layer Virtual Machine Resource Management," 9th International Conference on Autonomic Computing (ICAC), September 2012.
• Sajib Kundu, Raju Rangaswami, Ajay Gulati, Ming Zhao, Kaushik Dutta, "Modeling Virtualized Applications using Machine Learning Techniques," 8th Annual International Conference on Virtual Execution Environments (VEE 2012), March 2012.
• Lixi Wang, Jing Xu, Ming Zhao, Yicheng Tu, Jose Fortes, "Fuzzy Modeling based Resource Management for Virtualized Database Systems," 19th Annual Meeting of the IEEE International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems (MASCOTS 2011), July 2011.
• Yangyang Wu, Ming Zhao, "Performance Modeling of Virtual Machine Live Migration," 4th IEEE International Conference on Cloud Computing (CLOUD 2011), July 2011.
• Lixi Wang, Jing Xu, Ming Zhao, Jose Fortes, "Adaptive Virtual Resource Management with Fuzzy Model Predictive Control," 6th International Workshop on Feedback Control Implementation and Design in Computing Systems and Networks (FeBID, co-held with ICAC'11), June 2011.
• Lixi Wang, Jing Xu, Ming Zhao, Jose Fortes, "Adaptive Virtual Resource Management with Fuzzy Model Predictive Control," (Short Paper), 8th International Conference on Autonomic Computing (ICAC'11), June 2011.
• Dulcardo Arteaga, Ming Zhao, Chen Liu, Pollawat Thanarungroj, Lichen Weng, "Cooperative Virtual Machine Scheduling on Multi-core Multi-threading Systems — A Feasibility Study," Workshop on Micro Architectural Support for Virtualization, Data Center Computing, and Cloud (MASVDC, co-held with MICRO 2010), December 2010.
• Sajib Kundu, Raju Rangaswami, Kaushik Dutta, and Ming Zhao, "Application Performance Modeling in a Virtualized Environment," 16th IEEE International Symposium on High-Performance Computer Architecture (HPCA-16), January 2010.
• Jing Xu, Ming Zhao, José A. B. Fortes, Robert Carpenter, and Mazin Yousif, "Cooperative Autonomic Management in Dynamic Distributed Systems," 11th International Symposium on Stabilization, Safety, and Security of Distributed Systems (SSS 2009), November 2009.
• Lixi Wang, Jing Xu, Ming Zhao, Yicheng Tu, Jose Fortes, "Autonomic Resource Management for Virtualized Database Hosting Systems," Technical Report 2009-07-01, School of Computing and Information Sciences, Florida International University, July 2009.
• Jing Xu, Ming Zhao, and José A. B. Fortes, "Autonomic Resource Management in Virtualized Data Centers Using Fuzzy-logic-based Control," Cluster Computing, Vol. 11, No. 3, Pages: 213-227, September 2008.

Acknowledgement

This material is based upon work supported by the Department of Homeland Security under grant 2010-ST-062-000039 and a seed grant from Florida International University. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the sponsors.

---