透過您的圖書館登入
IP:18.116.8.110
  • 會議論文
  • OpenAccess

Efficient Resource Broker for Volunteer Workers Based on Adaptive Replicate Balancing

摘要


An opportunistic computing system typically employs replicas to obtain stable resources from unreliable workers. Usually, lifting the replication factor can increase the worker reliability but oppositely decrease the efficiency due to computation redundant. To address the best replication factor, many brokers adopt the unstable trial-and-error strategy. This paper develops a new resource broker for performing the prioritized malleable tasks over a pool of volunteer workers. The new approach controls the replication factor via the balancing on the system metrics that are positively or negatively related to the factor. With this balancing technique, the broker can rapidly control the replication factor to adaptive to the changing workload and thereby improves the system performance.

延伸閱讀