resource deployment method and system for a Linux container

Abstract

The invention provides a resource deployment method and system for a Linux container. The method comprises the steps that (1) for each physical host, the matching degree of the physical host and the container in each dimension of resource is calculated according to the requirement condition of the container to be built for N-dimensional resources and the condition that the physical host can provide the N-dimensional resources, and N is larger than 1; And 2) in the N-dimensional space, based on the feasible domain range determined by the set feasible domain boundary value m, one or more physical hosts of which the matching degrees is or are screened with the container on the resources of each dimension are in the feasible domain range, and m is greater than 0. According to the invention, resource fragments can be effectively reduced, the resource utilization rate of the host under the condition of close full load is improved, and resource waste is avoided. And moreover, compared with the traditional greedy strategy deployment, the resource balance degree of the method is improved by about 100% through testing.

Description

technical field [0001] The present invention relates to the field of computer technology, in particular to a resource deployment solution for Linux containers. Background technique [0002] As an important technology to realize the functions of the linux kernel, Linux container technology (LXC) has a development history of more than ten years. This technology isolates different processes from each other by building toolkits, and such isolation does not require hardware virtualization. reference here figure 1 Explain the principle of Linux container technology. As shown in the figure, the resources of the host machine (that is, the physical machine entity) are divided into different regions through the container as an isolation mechanism. Multiple containers share the same Linux kernel. Each container Accommodating one application makes the execution of programs for different applications independent of each other. It can be considered that the container is a layer of soft...

AI Technical Summary

Problems solved by technology

However, the resources in the cluster are usually multi-dimensional, such as CPU, memory, network bandwidth, GPU, block I / O bandwidth, storage, etc. If the balanced use of multi-dimensional resources is not guaranteed during resource deployment, it is easy to cause cluster failure. Resource fragmentation occurs in

Assuming that only the three-dimensional resources in the cluster, CPU, memory and storage are considered, based on the existing strategy, the CPU may run out and the memory still has a lot of memory, which will cause a waste of resources

Images