A dynamic load distribution strategy for systems under high task variation and heavy traffic
Proceedings of the 2003 ACM symposium on Applied computing
Value-maximizing deadline scheduling and its application to animation rendering
Proceedings of the seventeenth annual ACM symposium on Parallelism in algorithms and architectures
Review: Task assignment policies in distributed server systems: A survey
Journal of Network and Computer Applications
An extended evaluation of two-phase scheduling methods for animation rendering
JSSPP'05 Proceedings of the 11th international conference on Job Scheduling Strategies for Parallel Processing
An enhanced grid scheduling with job priority and equitable interval job distribution
GPC'06 Proceedings of the First international conference on Advances in Grid and Pervasive Computing
| Hi-index | 0.00 |
This paper describes a dynamic load balancing approach to distributed server farm systems. This approach overcomes the interference caused by non-negligible very-large tasks in the heavy-tailed distribution. First, a subset of tasks is allocated proportionally to the processing capability of participating servers by taking into account their remaining processing time. Later, tasks in the servers are processed in order of priority to optimise the system response time. The proposed load balancing algorithm also takes into account the information on server loads to avoid load imbalance caused by very large tasks. The experiments show that the mean waiting time and the mean slow down time are reduced at the server farm system.