Yarn architecture-based virtual machine scheduling system

Abstract

The present invention discloses a yarn architecture-based virtual machine scheduling system, and belongs to the field of cloud systems. The system comprises a Resource Manager module, a Node Manager module, a Yarn Proxy module, an Analysis Server module, and a Hypervisor host. According to the yarn architecture-based virtual machine scheduling system, function extension is performed on YARN, so that the system can simultaneously run a task scheduler and a virtual machine scheduler, and support the use of different scheduling algorithms to schedule different types of resources. According to the yarn architecture-based virtual machine scheduling system provided by the present invention, unified management is performed on resources, thereby greatly reducing operation and maintenance costs of different resources. The added Analysis Server module can analyze real-time resource usage of the virtual machine, and provide different oversold coefficients for users with different requirements, thereby solving a problem of low resource utilization.

Description

technical field [0001] The invention belongs to the field of cloud systems, in particular to a virtual machine scheduling system. Background technique [0002] In recent years, the tide of cloud computing has been surging, and now many companies, especially Internet companies, have established their own public or private clouds. The strong demand for cloud computing has promoted the development of virtualization technology. In order to improve resource utilization, virtualized servers have become the choice of most companies. However, the current mainstream virtual machine resource scheduling system can only meet specific needs and cannot be expanded. To other types of resource scheduling, unified management of system resources. The implementation of the scheduling algorithm is based on the amount of resources requested by the user. During the actual system operation, the virtual machine will not use the resources it occupies at full capacity most of the time, resulting in ...

AI Technical Summary

Problems solved by technology

The strong demand for cloud computing has promoted the development of virtualization technology. In order to improve resource utilization, virtualized servers have become the choice of most companies. However, the current mainstream virtual machine resource scheduling system can only meet specific needs and cannot be expanded. To other types of resource scheduling, unified management of system resources

The implementation of the scheduling algorithm used by the user is based on the amount of resources requested by the user. During the actual system operation, the virtual machine will not use its resources at full capacity most of the time, resulting in low utilization of system resources.

Moreover, for users with different needs, using the same standard to configure virtual machines may result in users with high performance requirements not being able to achieve smooth use. At the same time, ordinary users will also cause resource waste.

Images