摘要
为了提供具有可伸缩性的延迟保证,通过对虚拟时钟(GDVC)调度策略的研究,提出了核心无状态虚拟时钟(CS-GDVC)调度策略的框架,并证明它能够与基于流的虚拟时钟调度策略提供相同的端到端延迟保证。通过选择CS-GDVC中的一个参数和工作方式(工作守恒/不守恒),可以构造各种具体的核心无状态虚拟时钟调度策略。已经提出的各种基于虚拟时钟的核心无状态调度策略都可以认为是CS-GDVC的实例。CS-GDVC为核心无状态虚拟时钟调度策略的设计提供了完整的理论基础。
A general framework for corestateless virtual clock scheduling algorithms is proposed to provide scalable delay guarantees. The framework is based on the GDVC (generalized deferred virtual clock) scheduling algorithms, which provide the same endtoend delay bound as the per flow virtual clock scheduling algorithm. Various corestateless virtual clock scheduling algorithms can be built by selecting the parameter and the nature (workconserving or nonwork conserving) of the algorithm. All virtual clockbased corestateless scheduling algorithms can be shown to belong to this general framework which provides a complete theory for the design of virtual clockbased corestateless scheduling algorithms.
出处
《清华大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2003年第1期86-89,共4页
Journal of Tsinghua University(Science and Technology)
基金
国家自然科学基金资助项目(60173009)
清华大学"九八五"基金项目
