Login
Computer Science @ Rochester
Monday, January 05, 2009
11:00 AM
Computer Studies Bldg. Room 209
Cristiana Amza
University of Toronto
Dynamic Resource Allocation for Controlling Application Interference in Shared Servers
Due to the imperative need to reduce the costs of management, power and cooling in large data centers, operators multiplex several concurrent applications on each physical server of a server farm connected to a shared network attached storage. Determining and enforcing per-application resource quotas on the fly in this context poses a complex resource allocation and control problem within each physical server and/or across the server farm. This problem is further complicated by the need to provide end-to-end Quality of Service (QoS) guarantees to hosted applications.

In this presentation, I will introduce a novel multi-resource allocator to dynamically allocate resources for database servers running on virtual storage. Multi-resource allocation involves proportioning the client and storage caches, and the storage bandwidth between applications according to overall performance goals. The problem is challenging due to the interplay between different resources, e.g., changing any cache quota affects the access pattern at the cache/disk levels below it in the storage hierarchy. We use a combination of simulation, on-line modeling and sampling to arrive at near-optimal configurations within minutes. Our key ideas are to leverage known cache simulation and access tracking techniques as well as resource dependencies e.g., due to cache replacement policies, into our performance models.

Based on an experimental evaluation conducted on both micro-benchmarks and industry standard benchmarks, such as, TPC-W and TPC-C, we show that our multi-resource allocation approach provides up to a factor of 2.9 better overall performance compared to a combination of state-of-the-art single resource controllers.

Bio: Cristiana Amza is an Assistant Professor in the Department of Electrical and Computer Engineering at the University of Toronto. She received her Ph.D. and M.S. degrees in Computer Science from Rice University in 2003 and 1997, respectively, and her B.S. degree in Computer Engineering from Bucharest Polytechnic Institute in 1991. Her research interests are in the area of distributed and parallel systems, with an emphasis on self-adaptive behavior in dynamic content servers. She is currently investigating adaptation in a wide range of system components, from multi-core servers to mobile devices, and applications, from e-commerce to massively multiplayer games. She has established an autonomic computing laboratory providing cutting-edge research on large-scale distributed platforms. She is actively collaborating with several industry partners, including Bell Canada, IBM and Intel. Her awards include an Ontario Early Researcher Award (2007), an IBM Faculty Award (2006), several Intel Faculty Awards (2006-2008) and an Eclipse Award (2005).

Refreshments provided at 10:45AM