Supporting SLA And Dynamic Adaptive Cluster Management System Based On Hybrid Cloud

With the advent of container technology,container-based application deployment provides users with a new channel.With its integrated packaged operating environment for portability,users can easily create custom environments to suit the needs of their software products.Through the unified management of the multi-container scheduling and resource management by the cluster management system,it is necessary to have the ability to flexibly map the containers to different nodes and dynamically adjust container resources and migration containers according to load conditions.The agility with which these decisions can be made and implemented determines the responsiveness that enterprise IT can respond to changing business requirements,and has important theoretical value and broad application prospects for the research of cluster management systems that support container scheduling and management.The cluster management system studied in this paper mainly provides three major functions: application image resource allocation,automatic distribution and operation,container dynamic adaptation and hybrid cloud resource management when local resources are insufficient.Among them,resource allocation,automatic distribution,and operation of application image mainly implement automatic resource allocation and deployment operation of application images.Dynamic adaptation of the container mainly dynamically monitors the running load of the application within the container,and dynamically adds or deletes resources for the application;The cloud resource management mainly implements the dynamic apply the public cloud node for the deployment of application image when the resources of the local cluster are insufficient.When implementing resource allocation,automatic distribution and running of application images,in order to support the resource allocation of SLA,a formal structure of SLA is proposed to describe various types of resource requirements,and a dual-layer resource allocation algorithm is designed.To solve the problem of optimal resource allocation to the container when the number of types of requirements and computing resources is inconsistent.When dynamically adapting the container,some of the source code of the Run C engine is modified,and the iterative thermal migration of the container is achieved by using the network name space technology(Network Name Space)and port forwarding technology(Port Forwad),and at the same time,an iterative stop algorithm is proposed.The algorithm is used to reduce downtime during iterative migration.When solving the problem that container resources need to be stretched but local resources are insufficient for making migration decisions,proposing a cost comparison model that compares SLA default costs due to non-migrating performance degradation and service failure during downtime during migration.Comparison of SLA costs,the final decision whether to migrate.Design and implement section introduces the design and implementation of the entire system in detail.The architecture design of the whole system is introduced through the system architecture diagram and the functional module diagram.The design of the main functional modules is introduced through class diagrams,and the implementation logic of each functional module of the system is introduced through the sequence diagram.