2009 | OriginalPaper | Buchkapitel
Adaptive Scheduling for QoS Virtual Machines under Different Resource Allocation – Performance Effects and Predictability
verfasst von : Angela C. Sodan
Erschienen in: Job Scheduling Strategies for Parallel Processing
Verlag: Springer Berlin Heidelberg
Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by
Virtual machines have become an important approach to provide performance isolation and performance guarantees (QoS) on cluster servers and on many-core SMP servers. Many-core CPUs are a current trend in CPU design and require jobs to be parallel for exploitation of the performance potential. Very promising for batch job scheduling with virtual machines on both cluster servers and many-core SMP servers is adaptive scheduling which can adjust sizes of parallel jobs to consider different load situations and different resource availability. Then, the resource allocation and resource partitioning can be determined at virtual-machine level and be propagated down to the job sizes. The paper investigates job re-sizing and virtual-machine resizing, and the effects which the efficiency curve of the jobs has on the resulting performance. Additionally, the paper presents a simple, yet effective queuing-model approach for predicting performance under different resource allocation.