Docker dynamic scheduling method for typical containers

Abstract

The invention discloses a Docker dynamic scheduling algorithm for typical containers, including the following steps: S1: The scenarios of typical application containers include CPU-intensive / batch-processing, memory-intensive / batch-processing, I / O-intensive / batch Processing type and CPU-intensive / real-time type, which respectively select the corresponding application container, and analyze the resource usage and performance of each application container running independently or in multiple concurrent operations in the Docker environment; S2: Scheduling algorithm includes container static scheduling mode And the container dynamic scheduling method based on runtime monitoring; according to the user's needs, the container static scheduling method and the container dynamic scheduling method are adopted according to the situation. The beneficial effect of the present invention is that the dynamic scheduling algorithm can improve the utilization rate of system resources while not affecting the running performance of the application container.

Description

technical field [0001] The invention relates to the technical field of computer applications, in particular to a Docker dynamic scheduling algorithm for typical containers. Background technique [0002] Container technology is a lightweight alternative to virtual machines. While ensuring resource isolation between containers, its processing power, memory, and network throughput are close to the original performance of physical machines. As an application engine for containers, Docker can efficiently deploy, execute and manage containers. However, the existing Docker resource management mechanism is relatively simple, providing users with two methods: default resource configuration and manual configuration of container instance resources through parameters. However, there is no distinction between application container instance types, and the resource allocation for each type of container instance is relatively average. When real-time application containers and batch-proces...

AI Technical Summary

Problems solved by technology

However, the existing Docker resource management mechanism is relatively simple, providing users with two methods: default resource configuration and manual configuration of container instance resources through parameters.

However, there is no distinction between application container instance types, and the resource allocation for each container instance is relatively average

When real-time application containers and batch-processing application containers are running on the physical machine at the same time, it is difficult to quickly and dynamically adjust the resource configuration of the container instance according to the change of the service intensity of the real-time application container, so the service performance of the real-time application container cannot be guaranteed

[0003] Currently, Docker's existing resource management strategy does not perform resource limit checks based on the overall resource usage on the current physical machine, and does not limit the increase of container instances.

When multiple application containers with the same or similar resource types are running at the same time, the resource type used by the container is relatively single, which may easily lead to low utilization of other system resources; at the same time, due to resource competition, the resource requirements of the application container cannot be met, resulting in the container running poor performance

In addition, when the total memory resources used by the running container instances reach the system memory limit, the system may kill the normally running container due to insufficient current memory.

[0004] When a container instance is created and running, Docker provides users with a way to set the CPU share and disk I / O weight. However, if the user does not know the resource usage characteristics of the test application during use, it is difficult to determine the weight of resources. The type and the size of the corresponding parameter value

Images