Application Informed Tuning of Virtualized Environments
Virtualization is currently being used in cloud computing environments and traditional IT environments to improve the flexibility and manageability of the computing infrastructure, and to enable the sharing of computing resources. This means that applications (such as database systems) are increasingly being run on virtual machines and using virtualized storage. The performance of an application in this environment is affected by the configuration and tuning decisions made at the virtual machine/storage level. In this talk, I will demonstrate that coordinating between the application and the virtualization environment when making these tuning decisions can result in significant performance gains. I will present three examples of such application informed tuning: (1) configuring multiple virtual machines running database workloads on the same physical server, (2) improving the caching decisions of a storage server running a database workload, and (3) scheduling a batch of Map-Reduce jobs running on a cluster of virtual machines.
Ashraf Aboulnaga is an Assistant Professor in the David R. Cheriton School of Computer Science at the University of Waterloo. His research interests are in the area of database management, with a current focus on database issues in virtualized and cloud computing environments, self-managing database systems, XML databases, data integration, and data management on the web. Ashraf obtained his PhD from the University of Wisconsin in 2002. Prior to joining the University of Waterloo, he was a Research Staff Member in the Data Management Department at the IBM Almaden Research Center in San Jose, California, from 2002 to 2004. He is a holder of a Google Research Award in the area of data integration.
Speaker: Ashraf Aboulnaga, David R. Cheriton School of Computer Science, University of Waterloo
Google Tech Talk
April 16, 2009