摘要： 提出一种虚拟计算系统体系结构的自适应重构框架，以解决任务特征与体系结构的动态匹配问题。实现面向动态需求变化和面向物理资源变化的重构算法，在任务特征模型或底层物理结构变化时，能自适应进行线程及虚拟机的聚集，以形成新的任务分配和调度结构，使虚拟计算系统具有较高的运行效率，并通过虚拟机迁移，确保系统负载均衡。

关键词: 高效能计算, 任务特征模型, 虚拟计算系统体系结构, 自适应重构

Abstract: This paper presents a self-adaptive framework of Architecture of Virtual Computing System(AVCS) which provides an idea to solve the problem of applications not matching the architecture of AVCS. Two self-adaptive reconfiguration algorithms oriented the change of dynamic requirements from applications and physical resources are designed in this framework, which enable the threads gathering and virtual machines gathering automated and form a new task assignment and scheduling structure to improve the performance of virtual computing systems. Meanwhile, the virtual machines migration in the algorithms is implemented to keep the load balancing.

Key words: high performance computing, task feature model, Architecture of Virtual Computing System(AVCS), self-adaptive reconfiguration

中图分类号: 

• TP303