Design And Implementation Of Intelligent Pipe Gallery Container Cloud Platform Based On Kubernetes

In recent years,with the rapid development of container technology,the application of container technology to smart city has become a research hotspot.Based on the existing smart pipe gallery microservice architecture,this thesis designs a smart pipe gallery container cloud platform based on kubernetes container orchestration framework,and realizes the functions of container image simplification,multi-user level container management,monitoring and alarm service.At the same time,in order to deal with the time-consuming container scheduling in small and medium-sized container clusters and the impact on service performance,this thesis proposes a container scheduler design and implementation scheme based on ant colony algorithm from the aspects of container scheduling queue and parallel scheduling algorithm.The main work is as follows:(1)Aiming at the problem that the available resources of cloud servers are limited and the traditional open source containers occupy more resources,which leads to the loss of system performance,this thesis proposed a lightweight container construction scheme for the pipe gallery system.In order to reduce the consumption of server resources occupied by the image and container,the method reduced the dependence of each layer image and constructs the underlying common image.The test results show that this method can greatly reduce the server resources occupied by ordinary images,improve the resource utilization efficiency and improve the system performance.(2)In order to improve the reliability and maintainability of smart pipe gallery container system,combined with Prometheus and grafana,a service monitoring and alarm architecture of pipe rack is designed,and the functions of container monitoring,alarm and monitoring index visualization are realized.The scheme combines the original dashboard of kubernetes as a platform to build a new multi-user level container management monitoring alarm system.Compared with the mainstream monitoring architecture,it not only improves the multi-user level management and monitoring level of the container,but also can send an alarm to the operation and maintenance personnel to ensure the reliable operation of the service in real time.(3)Aiming at the scheduling time-consuming problem of native kubernetes scheduler in small and medium-sized clusters with low cluster heterogeneity and few container service types,this thesis proposed a design and implementation scheme of kubernetes container cluster scheduler based on ant colony algorithm.Based on the original priority queue,the scheduler optimized the operation mode of the queue to pod,and adopted the optimal ant colony algorithm as the main scheduling mode of container scheduling.At the same time,combined with the complex scheduling strategy that may appear in the cluster,the original scheduling method is used as a backup scheme,which improved the fault tolerance and reliability of the scheduler.Theoretical analysis and test results show that the proposed scheduler design and implementation scheme can significantly shorten the scheduling time and improve the system performance compared with the current mainstream scheduling algorithm in the industry.