Research And Implementation Of Multi-cluster Edge Cloud Scheduling Mechanism In Edge Environment

Edge computing is a computing model that sinks computing and storage to the edge,and multi-cluster edge cloud is one of the mainstream models of today’s edge computing architecture.The ability to achieve reasonable scheduling of data and computing resources in a multi-cluster edge cloud environment is a core element that determines the execution efficiency of edge computing.The current scheduling method has problems such as unbalanced cluster scheduling load and large task scheduling delay.For this reason,this thesis proposes a multi-cluster edge cloud scheduling strategy based on the different task requirements in the edge environment.details as follows:First of all,this thesis designs a new multi-cluster edge cloud framework,and designs some functional modules and related workflows on this framework to realize task scheduling and distribution.The framework divides the tasks into delay-sensitive and insensitive tasks according to their different requirements for edge nodes,and adopts different processing methods for them.Aiming at the high requirements of delay-sensitive tasks for delay,a distribution strategy for delay-sensitive tasks in a cluster is designed.The specific process of the strategy is as follows: first query the remaining processing capacity of each node,and then use the improved ant colony algorithm according to the situation of the node,take the task’s resource demand as the heuristic,and modify the pheromone update rules,Reasonably allocate tasks with different requirements on the nodes to achieve overall load balancing of the cluster.For delay-insensitive tasks,their focus is on data desensitization and cleaning,and the requirements for delay are not high.In this thesisa cross cluster task scheduling strategy is proposed,which can complete the cross cluster scheduling function of delay insensitive tasks and improve the utilization of cluster resources by scheduling trigger time,scheduling task selection,scheduling cluster selection and other aspects.Finally,the large-scale data verification carried out in the constructed edge cluster cloud environment,compared with the existing scheduling method under the edge computing environment,the overall waiting time of the task is reduced by 10.2%,the service quality of the cluster is higher,and the overall balance is also better.