基于多任务拍卖的资源调度算法

Resource Scheduling Algorithm Based on Multi-Job Auction

资源调度是计算网格资源管理系统的主要内容之一。文中借鉴市场模型中的拍卖机制,通过用户代理、资源代理、拍卖师和资源的交互作用,设计并实现了一个基于多任务拍卖的网格资源调度原型系统,系统使用多拍卖师结构,提高了作业的吞吐率并且可以有效地防止拍卖过程中双方的欺骗行为,改进了网格资源的调度性能。同时,与传统算法相比,该算法能够有效地配置资源和满足用户的服务质量需求。

Resource scheduling is the most important problem in the computational grid system. A resource scheduling prototype system of multi - job architecture based on computing econoray model, because of the interaction among the user - agents, resource- agents, auctioneers and resources, is designed and implemented. The scheduling system which adopts the multi - auctioneer increases the throughput of jobs for computational grid system. On the other hand,it prevents from the fraud action of the two parties. It improves the performance of grid resource, scheduling system .And in comparison with the conventional algorithms, it can configure resource more efficaciously as well as meet the requirements of the users＇ QoS better.