Virtualized container service elastic scaling method oriented to data center load monitoring

Abstract

The invention provides a virtualized container service elastic scaling method oriented to data center load monitoring. Within the cloud platform cluster, at first, the real-time load of the containerservice is obtained through the load monitor service, and according to the load and the relationship of the upper bound and the lower bound of the set threshold, whether the elastic scaling of the container service is triggered is determined, after detecting that the container load pressure exceeds the upper limit of the threshold, container service expansion is triggered. By synthetically scoringdifferent nodes in the cluster with various algorithms, the best adapter node for elastic expansion and contraction is obtained, and finally a new container is started on the adapter node to completethe container service expansion. As the load pressure of the container is detected to be low than the lower threshold limit, the container service shrinkage is triggered, and elastic expansion and contraction of the container service are completed by removing the corresponding container on the node marked in the expansion phase. The method can realize the elastic expansion of cluster inner container service on the premise of keeping the load balance of cluster resources.

Description

technical field [0001] The invention relates to resource scheduling of cloud platform clusters, in particular to a virtualized container service elastic scaling method for data center load monitoring, which belongs to the field of computer science and technology. Background technique [0002] As a product of the development and fusion of traditional computer and network technologies such as distributed computing, virtualization, grid computing, and network storage, cloud computing has become one of the most influential changes in global information technology. , also profoundly influenced and changed people's way of life. Cloud computing is a computing method that shares software and hardware resources and information through the Internet. It provides IT-related capabilities to users in the form of services, so that users do not understand the technology that provides services, have no relevant knowledge and equipment operating capabilities. It realizes fast and on-demand u...

AI Technical Summary

Problems solved by technology

[0004] Cluster node load imbalance: The traditional load-triggered random scheduling scaling method does not consider the status of nodes in the cluster and resource distribution when scaling is triggered, which may lead to severe unbalanced cluster resource load, resulting in a single-point bottleneck of cluster node resource utilization

Images