The GW Cloud Systems Lab's research projects span a range of challenges faced by cloud computing data centers. Throughout all of our work, our focus is on developing new technologies and control algorithms that can be prototyped and deployed in next generation cloud platforms.
This page covers only a small portion of the work being done in our lab. Please see our Publications page for more details on our work.
To see our open source code releases, check out our github organization.
Cloud Resource Management
Cloud platforms rely on densely packed data centers that are geographically distributed across the world. Our work has developed both new virtualization-layer mechanisms to manage resources and new machine-learning based control algorithms to automate cloud management.
- Matrix: Achieving Predictable Virtual Machine Performance in the Clouds. Ron C. Chiang, Jinho Hwang, Howie Huang, Timothy Wood. USENIX International Conference on Autonomic Computing (ICAC), June 2014.
- Mortar: Filling the Gaps in Data Center Memory. Jinho Hwang, Ashen Uppal, Timothy Wood, Howie Huang. International Conference on Virtual Execution Environments (VEE), March 2014.
- GPU-Accelerated Scalable Solver for Banded Linear Systems. Hang Liu, Jung Seo, Rajat Mittal, Howie Huang. IEEE Cluster 2013 (Cluster’13), Indianapolis, Indiana, September 2013.
- Adaptive Dynamic Priority Scheduling for Virtual Desktop Infrastructures. Jinho Hwang, Timothy Wood. IEEE / ACM International Workshop on Quality of Service (IWQoS), June 2012.
- Energy-Aware Writes to Non-Volatile Main Memory. Jie Chen, Ron Chiang, Howie Huang, Guru Venkataramani. 4th Workshop on Power-Aware Computing and Systems (HotPower’11), with the 23nd ACM Symposium on Operating Systems Principles (SOSP'11), October 2011.
- TRACON: Interference-aware Scheduling for Data-intensive Applications in Virtualized Environments. Ron C. Chiang, H. Howie Huang. Proceedings of 2011 International Conference for High Performance Computing, Networking, Storage and Analysis, 2011.
Storage Systems & Data Intensive Computing
Data centers can contain petabytes of storage, which must be efficiently managed if data-intensive applications wish to access it. Our work seeks to optimize virtual clusters to increase the performance of big data applications.
- UniCache: Hypervisor Managed Data Storage in RAM and Flash. Jinho Hwang, Wei Zhang, Ron C. Chiang, Timothy Wood, Howie Huang. IEEE International Conference on Cloud Computing (CLOUD), June 2014.
- MIMP: Deadline and Interference Aware Scheduling of Hadoop Virtual Machines. Wei Zhang, Sundaresan Rajasekaran, Timothy Wood, Mingfa Zhu. IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing, May 2014.
- HybridMR: A Hierarchical MapReduce Scheduler for Hybrid Data Centers. Bikash Sharma, Timothy Wood, Chita R. Das. IEEE International Conference on Distributed Computing Systems (ICDCS 2013), July 2013.
- Flashy Prefetching for High-Performance Flash Drives. Ashen Uppal, Ron Chiang, Howie Huang. IEEE Conference on Mass Storage Systems and Technologies (MSST’12)., April 2012.
- Performance Modeling and Analysis of Flash-based Storage Devices. Howie Huang, Shan Li, Alexander Szalay, Andreas Terzis. IEEE Symposium on Massive Storage Systems and Technologies (MSST), May 2011.
- Just-In-Time Analytics on Large File Systems. Howie Huang, Nan Zhang, Wei Wang, Gautam Das, Alexander Szalay. Proceedings of the 9th USENIX Conference on File and Storage Technologies (FAST) , February 2011.
While performance and cost are the most obvious metrics in the cloud, reliability can be the most important factor for many businesses. We aim to increase the reliability and fault tolerance of the underlying software and hardware systems in the cloud.
- Understanding Reliability Implication of Hardware Error in Virtualization Infrastructure. Xin Xu, H. Howie Huang. 10th Workshop on Hot Topics in System Dependability (HotDep 14), October 2014.
- Xentry: Hypervisor-Level Soft Error Detection. Xin Xu, Ron Chiang, H. Howie Huang. The 43rd International Conference on Parallel Processing (ICPP14), September 2014.
- Exploring Dynamic Redundancy to Resuscitate Faulty PCM Blocks. Jie Chen, Guru Venkataramani, Howie Huang. ACM Journal of Emerging Technologies in Computing Systems, 2014.
- Exploring Data-Level Error Tolerance in High-Performance Solid-State Drives. Xin Xu, Howie Huang. IEEE Transactions on Reliability, 2014.
- DUAL: Reliability-Aware Power Management in Data Centers. Xin Xu, Kayo Teramoto, Allan Morales, Howie Huang. IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGrid’13), Delft, the Netherlands, May 2013.
- Providing Reliability as an Elastic Service in Cloud Computing. Nakharin Limrungsi, Juzi Zhao, Yu Xiang, Tian Lan, Howie Huang, Suresh Subramaniam. IEEE International Conference on Communications (ICC’12), June 2012.
- RePRAM: Re-cycling PRAM Faulty Blocks for Extended Lifetime. Jie Chen, Guru Venkataramani, Howie Huang. International Conference on Dependable Systems and Networks (DSN), June 2012.
- PipeCloud: Using Causality to Overcome Speed-of-Light Delays in Cloud-Based Disaster Recovery. Timothy Wood, Andres Lagar-Cavilla, K. K. Ramakrishnan, Prashant Shenoy, Jacobus Van der Merwe. 2nd Symposium on Cloud Computing (SOCC), October 2011.
Optimizing the networks within these data centers, while still granting control to cloud customers remains a major challenge. Of course, efficiently connecting and managing data center across wide area networks also poses many challenges. We are developing infrastructure management and monitoring technologies, as well as entirely new approaches for running network services inside virtual machines.
- Load Balancing of Heterogeneous Workloads in Memcached Clusters. Wei Zhang, Timothy Wood, H. Howie Huang, Jinho Hwang, K.K. Ramakrishnan. Usenix International Workshop on Feedback Computing, June 2014.
- SmartSwitch: Blurring the Line Between Network Infrastructure & Cloud Applications. Wei Zhang, Timothy Wood, K.K. Ramakrishnan, Jinho Hwang. Usenix Workshop on Hot Topics in Cloud Computing (HotCloud), June 2014.
- NetVM: High Performance and Flexible Networking using Virtualization on Commodity Platforms. Jinho Hwang, K.K. Ramakrishnan, Timothy Wood. Symposium on Networked System Design and Implementation (NSDI), April 2014.
- Topology Discovery & Service Classification for Distributed-Aware Clouds. Jinho Hwang, Guyue Liu, Sai Zeng, Frederick y Wu, Timothy Wood. IEEE International Workshop on Cloud Analytics (IWCA14), March 2014.