Container dynamic balance scheduling method

Abstract

The invention relates to the technical field of containers, in particular to a container dynamic balance scheduling method, which comprises the following steps that: a scheduler monitors newly created container group resources, and schedules containers to proper nodes through custom scheduling assertion and strategy; the equalization scheduler monitors the utilization rate of node resources, and when the node resources are unbalanced, the node container groups occupying more resources are expelled; the batch controller monitors the job resources and creates container group resources; after monitoring the container group resources, the batch scheduler performs resource scheduling by taking the container group as a unit; and the container manager performs life cycle management of the container groups according to the container group binding node information, thereby expelling the container groups with low priorities or wasting more resources according to the expelling configuration when the node resources are insufficient. Compared with the prior art, the method has the advantages that the resource scheduling performance of the container cloud is improved, the probability of expelling key containers is reduced, and the high availability of the container cloud is ensured.

Description

technical field [0001] The present invention relates to the technical field of containers, and specifically provides a container dynamic balance scheduling method. Background technique [0002] Container technology is a popular PaaS technology that provides microservices, and Kubernetes is the de facto standard for container orchestration and scheduling. Providing container cloud platforms based on Kubernetes is the first choice of major cloud vendors. Kubernetes provides default scheduling and eviction algorithms. When a container group is being scheduled, the container group is scheduled to the corresponding node according to these scheduling algorithms. [0003] The assertions and policies of container scheduling can be customized, for example, by defining the number of nodes for each scheduling query to improve the scheduling speed, by modifying the loading order and weight of assertions and policies, and customizing container group scheduling actions to meet the schedul...

AI Technical Summary

Problems solved by technology

[0006] When a node is abnormally restarted or a new node is expanded, Kubernetes will not automatically balance the node container group with high load to the node with low load, which will cause load imbalance among cluster nodes. There is an urgent need for a scheduling algorithm that dynamically balances each node.

Images