PERFORMANCE ISOLATION - SHARING AND ISOLATION IN SHARED-MEMORY MULTIPROCESSORS

Citation
B. Verghese et al., PERFORMANCE ISOLATION - SHARING AND ISOLATION IN SHARED-MEMORY MULTIPROCESSORS, ACM SIGPLAN NOTICES, 33(11), 1998, pp. 181-192
Citations number
25
Categorie Soggetti
Computer Science Software Graphycs Programming","Computer Science Software Graphycs Programming
Journal title
Volume
33
Issue
11
Year of publication
1998
Supplement
S
Pages
181 - 192
Database
ISI
SICI code
Abstract
Shared-memory multiprocessors (SMPs) are being extensively used as gen eral-purpose servers. The tight coupling of multiple processors, memor y, and I/O provides enormous computing power in a single system, and e nables the efficient sharing of these resources. The operating systems for these machines (UNIX or Windows NT) provide very few controls for sharing the resources of the system among the active tasks or users. This unconstrained sharing model is a serious limitation for a server because the load placed by one user can adversely affect other users' performance in an unpredictable manner. We show that this lack of isol ation is caused by the resource allocation scheme (or lack thereof) ca rried over from single-user workstations. Multi-user multiprocessor sy stems require more sophisticated resource management, and we show how the proposed ''performance isolation'' scheme can address the current weaknesses of these systems. We have implemented performance isolation in the Silicon Graphics IRIX operating system for three important sys tem resources: CPU time, memory, and disk bandwidth. Running a number of workloads we show that our proposed scheme is successful at providi ng workstation-like isolation under heavy load, SMP-like latency under light load, and SMP-like throughput in all cases.