一种改进的基于动态信任的分布式调度模型

An Improved Distributed Scheduling Model Based on Dynamic Trust

任务调度是分布式计算系统研究的核心内容之一,而基于动态信任的任务调度算法在志愿计算和网格计算系统中具有很好的性能,这种算法能够较好地处理节点的不确定性。首先介绍了现有的基于动态信任的任务调度模型,分析了经典调度模型中没有考虑任务大小、执行期限和计算冗余组计算能力等缺点。在经典模型的基础之上,提出了一种在BOINC系统中适用于独立计算任务的基于动态信任的分布式任务调度改进模型。该模型对经典模型中可靠度的定义作了优化,并且考虑了任务大小、执行期限、计算冗余组计算能力等因素,从而提高了计算的吞吐量,减少了计算时间。

Task scheduling is one of the core contents in distributed computing system study area. The task scheduling model based on dynamic trust has been confirmed to have good performance in volunteer computing and grid computing systems, which can deal well with the uncertainty of the node. This paper firstly introduces the general situation of the existing task scheduling model based on dynamic trust, and analyzes the shortages of the model, such as task size, deadline and computing power of the redundancy group. Then a new task scheduling model is proposed, which can be used in independent tasks calculation of BOINC. The new algorithm optimizes the definition of reliability and takes the size and deadline of task and the calculation capability of the redundant group into account, which improves the calculation throughput and reduces the computation time.