An examination of strategies for estimating capacity to share among private workstations
ACM SIGSMALL/PC Notes
Future Generation Computer Systems - Special issue on metacomputing
Homeostatic and Tendency-Based CPU Load Predictions
IPDPS '03 Proceedings of the 17th International Symposium on Parallel and Distributed Processing
Finding Idle Work Periods on Networks of Workstation
Finding Idle Work Periods on Networks of Workstation
Resampling Methods: A Practical Guide to Data Analysis
Resampling Methods: A Practical Guide to Data Analysis
High-end workstation compute farms using windows NT
WINSYM'99 Proceedings of the 3rd conference on USENIX Windows NT Symposium - Volume 3
| Hi-index | 0.00 |
The basic idea of grid computing is a better use of underutilized resources. Following this idea, desktop grids target ordinary workstations, which are very powerful today. However, due to the priority of the local users, it is impossible to exclusively reserve com puting time for grid jobs on these machines. Consequently, an already running grid job might be delayed or even canceled. A forecast of future available computing capacities could alleviate this problem. Such a prediction would be especially useful for the allocation of the most appro priate machines and for making stochastic assertions on the completion of submitted jobs. In this paper we discuss suitable approaches for predicting the availability of computer resources. We develop measures to finally make a comparison of the approaches, which is based on empirical data from available workstations.