面向分布式环境的信号驱动任务调度算法

Task scheduling algorithm for distributed environment based on signal-driven

为优化Iaa S服务的执行效率,提出面向Iaa S的信号驱动任务调度算法,该算法根据Iaa S模型的结构特征建立控制子系统和节点子系统,根据任务的结构特征建立任务的DAG(directed acyclic graph)调度模型,并建立各任务分片的状态转化机制及控制子系统和节点子系统间的信号通信机制。以系统间信号交互的方式驱动任务分片的状态改变,并在每一调度时刻来临时利用并行优化选择策略分配任务分片。由于本算法采用了模拟Iaa S模型的双系统控制方式,使本算法与Iaa S模型的分布式体系相兼容且复杂度较低。最后通过实验验证了所提算法的有效性和实用性。

In order to optimize the performance of user services in Iaa S, the task scheduling algorithm for Iaa S based on signal-driven is proposed, by which CS（control subsystem） and NS（inquiry nodes subsystem） based on the structural characteristics of the Iaa S is established, and the DAG scheduling model based on the structural characteristics of the inquiry task is created. Then the conversion mechanism for the task partitions is created, constructing the signal communication mechanism for CS and NS, changing the status of the task partitions by signal-driven between the CS and NS, completing the task partitions allocation by POSS（parallel optimization selective strategy） in the scheduling time. This algorithm with low complexity is compatible with the distributed architecture of Iaa S, because of utilizing dual system control mode. The effectiveness and practicality of this algorithm is verified by experiment.