Dynamic allocation method for container resources in cluster

Abstract

The invention discloses a dynamic allocation method for container resources in a cluster; wherein the method comprises the steps: firstly carrying out the analysis of a historical resource use time sequence of a container through an LSTM-RBF model, and predicting the future resource use amount of the container; secondly, establishing a multi-container multi-resource allocation model, wherein different types of applications are considered in the model so that different weights are set for the applications with different QoS requirements, and the weight can be set according to the requirement degree of each application for each resource; then, establishing a system utility maximization model, so that and multiple resources used by the multiple containers are dynamically allocated with the goal of maximizing the overall utility of the system; and finally, solving a multi-container multi-resource problem by adopting an improved whale optimization algorithm. Various resources used by a plurality of containers are dynamically scheduled simultaneously from the perspective of the whole system, so that the maximization of the system utility can be achieved, and the utilization rate of the whole system resources is optimal.

Description

technical field [0001] The invention belongs to the technical field of computer container clusters, and in particular relates to a method for dynamically allocating container resources in a cluster. Background technique [0002] With the rapid development of container virtualization technology represented by docker, its advantages such as low overhead, fast startup, easy deployment, and application-centricity make it widely used in application scenarios such as cloud computing and edge computing. Based on container virtualization technology And container cluster management tools (such as k8s) provide services to users on demand. However, in a cluster environment containing multiple container applications, how to dynamically allocate multiple resources to containers on demand has not been well resolved. [0003] There are mainly two types of existing container resource allocation methods: static and dynamic. At present, the most popular k8s cluster basically adopts the meth...

AI Technical Summary

Problems solved by technology

At present, the most popular k8s cluster basically adopts the method of static allocation and manual modification. When creating a container, the user specifies the required resources through the limit and request parameters or allocates certain resources by default when the user does not specify. When the resource requirements of the container change When manually expanding the container horizontally or vertically, the static allocation strategy cannot meet the real-time changing resource requirements of the container. At the same time, it also allows users to set QoS. When resources are tight, K8S will kill low-priority containers to ensure the resource usage of high-priority containers. Requirements; dynamic resource allocation methods generally use specific models to predict the future resource usage of containers, and dynamically adjust container resources or select appropriate nodes for container deployment in combination with container resource usage restrictions and system available resources.

However, the existing container resource dynamic allocation strategy only dynamically adjusts the resources of the concerned target container or selects the appropriate node for deployment. The resources of other containers in the system will not be dynamically adjusted, and there is no simultaneous use of multiple containers from the perspective of the system as a whole. It is difficult to achieve the optimization of the overall resource utilization of the system through dynamic scheduling of various resources.

Images