A fair and efficient virtual machine scheduling system and method

Abstract

The invention discloses a virtual machine scheduling system and method with fairness and efficiency balance. The task request of the server is forwarded to the core module of the scheduler; the core module of the scheduler is connected with the cluster resource monitoring module, the task request processing module and the virtual machine opening module respectively, and the scheduler core module obtains information from the cluster resource monitoring and task request processing module, After the processing is completed, it is passed to the virtual machine start module, and the virtual machine is created to the physical node, and finally the virtual machine scheduling process is completed. The present invention comprehensively considers the fairness of allocation and the task communication requirements, allocates resources according to the principle of fairness, and merges the states of the intermediate results of allocation according to the task communication requirements, so as to optimize the allocation results, obtain more reasonable configurations, and realize task execution Comprehensive optimization of efficiency and cluster load.

Description

technical field [0001] The present invention relates to the field of computer science and technology, and further relates to the field of IaaS (Infrastructure as a Service, infrastructure as a service) in cloud computing, specifically a virtual machine scheduling method with fairness and efficiency balance. The invention can be used in the scheduler component of the cloud computing platform to provide more reasonable scheduling strategy support for the cloud computing platform and obtain more reasonable virtual machine allocation configuration. Background technique [0002] The continuous development of computing technology and the diversification of user application requirements make application providers provide more and more diverse solutions, which will cause waste of resources and overhead caused by adapting to different platforms. Cloud computing uses virtualization technology to reorganize and virtualize the underlying physical resources to form a huge resource pool. ...

AI Technical Summary

Problems solved by technology

The scheduler reduces the possibility of efficient collaboration between virtual machines with collaborative requirements, thus affecting the quality of service of multiple large-scale applications that require collaborative computing

[0013] 2) Maximizing cluster utilization reduces the number of physical machines used, but the physical cluster runs under partial load. If a physical node goes down, it will cause a large number of virtual machines to go down

[0014] 3) Random allocation increases the uncontrollability of the control part, and the scheduler cannot associate allocation, which reduces the possibility of efficient coordination

Due to the characteristics of random behavior, it is easy to cause some extreme situations, resulting in low efficiency of system operation

[0015] 4) Dynamic migration technology brings dynamic features to scheduling, but it also brings increased scheduling costs, such as the preservation of the instantaneous running state of the virtual machine, the migration of virtual mirror copies, etc.

[0016] To sum up, the current typical virtual machine scheduling methods only consider static policy configuration, single host utilization, cluster load, etc., while ignoring the attributes of tasks, which reduces the execution efficiency of tasks

Images