 |
Center for Nonlinear Studies |
High-performance computer systems continue to grow in size and complexity and must frequently handle many different jobs simultaneously. The I/O subsystem is often a bottleneck to overall system performance and interference among applications can lead to disproportionate performance degradation, unpredictable run times, and inefficient use of the resources. This talk presents our ongoing research in managing and guaranteeing, the performance of large, distributed storage systems. We will discuss our general model for performance management, survey our solutions for CPU, disk, network, and storage server cache, and discuss our research toward applying these solutions to the monitoring and management of distributed systems.
Bio:
Dr. Scott A. Brandt is Professor of Computer Science at the University of California, Santa Cruz and UCSC Director of the UCSC/Los Alamos
Institute for Scalable Scientific Data Management (ISSDM). His interests are broadly in the are of Computer Systems and his research ranges from scalable, high-performance, distributed storage to real-time CPU scheduling. He received his Ph.D. in Computer Science from the University of Colorado at Boulder in 1999.
Host: Sarah Michalak, CCS-6