基于结构化对等网络的高效分布式任务调度

摘要

A structure P2P approach for High-Efficient Distributed Task Scheduling policy called HDTS is presented and described. HTDS is Peer-to-Peer based, decentralized, scalable, and large-scale, these features are strongly supported by a structured P2P network overlay over which peers can communicate with each other directly and bi-directionally. Chord style ring addresses some important issues regarding self-organization and efficiency. HTDS allows any P2P node to initiate distributed computations which get efficiently distributed subtask among participating nodes. Plus, jobs comprised of communicating tasks are also supported as required for example in travelling salesman problem. The experimental results obtained from performance analysis show that HTDS is feasible and efficient approach which can provide a new way for computing resource sharing and aggregation.%提出与描述了一个基于结构化对等网络的高效分布式任务调度策略HDTS (High-Efficient Distributed Task Scheduling).HDTS建立在结构化对等网络的应用层覆盖网络上,保证了系统的非集中性、可扩展性、自组织性,以及规模大的优点.HDTS把基于Chord风格的对等网络协议和容错及高效的多播调度策略结合起来,允许分布式计算的各种大量的子任务在对等网络的节点上高效的调度、分配、执行.除了支持主-从风格的并行计算外,系统允许具有数据依赖的分布式旅行商算法正确的执行,使系统具有通用性和开放性.测试结果表明:HDTS具有正确性与高效性,可以作为对等网络上计算资源高效共享与聚集的可行方案.