Design And Implementation Of Resource Scheduling And Docker-based Application Container Cluster Management System

With the growing maturity of Docker virtualization technology,the application development,deployment,operation and maintenance of Docker-based application container cluster are gradually becoming realistic choice for more and more business and service operators.How to better realize the efficient use of enterprises' IT resources and to provide a highly available supporting environment for complex distributed applications deployed in Docker application container cluster are urgent problems to be solved for enterprises' IT resource management.To meet the above needs,variety of Docker application container cluster management tools with open resource,with Kubernetes and Swarm as representatives,are introduced in the industry and gains wide attention.However,such management tools like Kubemetes still have many deficiencies.Firstly,functions like application update and application migration provided by them can only be completed after closing the application program operation,which fails to provide 24/7 persistent service for users.Secondly,resource scheduling algorithm can only be achieved in accordance with preset rules.Such static scheduling cannot satisfy the dynamic adjustment need in consideration of limited resources.Thirdly,IT resource management of such platform needs operation and maintenance staff to provide more manual interventions,which may increase operation and maintenance staff s burdens.Therefore,it's necessary to develop Docker-based resource dynamic scheduling and application container cluster management system that adapts to enterprises' application status in order to better satisfy enterprises' needs to efficient use of IT resources and provide high-reliable services with for users.Based on the above background,this paper will give a comprehensive introduction about Docker-based resource scheduling and application container cluster management system in such aspects as research and development target,demand orientation,systematic design,implementation,and testing.To solve problems like the unsatisfactory effect of resource scheduling and redistribution as well as low resource utilization rate existing in the current resource scheduling methods,the paper provides an improved SLA-driven resource scheduling method to realize load balancing and dynamic adjustment of IT resource.As for problems existing in the current application container cluster management tools,it is expected to conduct researches in incremental update of application container,Hazelcast ram grid,as well as discovery and registration of network and service,with combination of Kubernetes application container cluster management framework.Adopting B/S structure and MVC design pattern and utilizing Node.JS,a proven Web application developer network,a Docker-based resource scheduling and application container management system is designed to effectively solve problems involving shortcut automatic deployment of application and service,smooth application update and smooth application migration.The main functional modules of the system include application packages construction,resource monitoring,resource scheduling,application automatic deployment,automatic application update,and application migration.Currently,the system has been put into trial operation.Test results show that the system can better satisfy the application requirement of enterprises' IT resource management.Under such circumstance that large numbers of concurrent users access to the system,the resource scheduling of the system can provide sound guarantee for user services and also give reliable support for enterprises' work such as application development,delivery,deployment,and update management.