DMCA
Using runtime measured workload characteristics in parallel processor scheduling (1996)
Cached
Download Links
| Venue: | University of Washington |
| Citations: | 32 - 2 self |
BibTeX
@INPROCEEDINGS{Nguyen96usingruntime,
author = {Thu D. Nguyen and Raj Vaswani and John Zahorjan},
title = {Using runtime measured workload characteristics in parallel processor scheduling},
booktitle = {University of Washington},
year = {1996},
pages = {155--174},
publisher = {Springer-Verlag}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
We consider the use of runtime measured workload characteristics in parallel processor scheduling. Although many researchers have considered the use of application characteristics in this domain, most of this work has assumed that such information is available apriori. In contrast, we propose and evaluate experimentally dynamic processor allocation policies that rely on determining job characteristics at runtime; in particular, we focus on measuring and using job efficiency and speedup. Our work is intended to be a first step towards the eventual development of production schedulers that use runtime measured workload characteristics in making their decisions. We consider two distinct scheduling scenarios: interactive systems, where minimizing response time is the goal, and batch systems, where maximizing useful instruction throughput is the goal. In both these environments, our experimental results validate the following observations: • Despite the inherent inaccuracies of runtime measurements and the added overhead of more frequent reallocations, schedulers that use runtime measurements of workload characteristics can significantly outperform schedulers that are oblivious to these characteristics. • Runtime measurements are sufficient for schedulers to achieve performance surprisingly close to that possible when efficiency and speedup information is available apriori. • The primary performance loss, relative to the use of aprioriinformation, is due to the transient decisions of the schedulers as they acquire information on the running applications, rather than to measurement and reallocation overheads. Our experiments are performed using prototype implementations running on a 50-node KSR-2 shared memory multiprocessor.
Keyphrases
workload characteristic parallel processor runtime measurement application characteristic added overhead following observation transient decision inherent inaccuracy prototype implementation memory multiprocessor eventual development 50-node ksr-2 dynamic processor allocation policy many researcher speedup information job characteristic useful instruction throughput response time frequent reallocation batch system job efficiency primary performance loss first step available apriori production scheduler experimental result running application interactive system
