Research And Implementation Of Docker Container Scheduling Strategy In Micro Service

The micro service architecture is a technology that deploys applications and services in the cloud,which decouples applications into multiple sub-modules based on their responsibilities.Each sub-module is deployed independently and interacts through a lightweight communication mechanism,which solves the growing user demand for system performance.The deployment of the micro service architecture can be packaged together with the system operating environment through container technology,it greatly reduces the difficulty of system operation and maintenance.However,due to smaller granularity of the container compared to the traditional virtual machine,the resource scheduling problem is also more complicated.In the actual application scenario,the container cannot be scheduled and managed well without manual intervention.And as the size of the container continues to increase,the relationship between containers becomes more complex and the management complexity also increases.In this case,the overall performance of the system and the correctness of manual operations are difficult to guarantee.How to efficiently manage the container is a problem that needs to be solved in the micro service environment.The main work of this paper is as follows:1.A workload prediction algorithm based on Gated Neural Units(GRUWP)is proposed.The workload model of the GRUWP algorithm fully considers the impact of memory,CPU,disk IO,and network IO on the workload.Also,the algorithm uses the GRU neural network model to predict the workload.Experimental results show that the workload prediction algorithm proposed in this paper can not only ensure a higher prediction accuracy,but also has higher computational efficiency than the existing workload prediction algorithm based on neural networks.2.Study and analyze the impact of dependencies between containers on the performance of micro services.The containers are divided into neighborhoods based on their dependencies and the neighborhoods are integrated into the key parameters of the particle swarm optimization algorithm.According to the above research results,a container scheduling algorithm based on neighborhood division is proposed.Simulation experiments show that the scheduling algorithm can not only ensure a certain workload balance,but also reduce the network calls across the physical machine.3.Design and implement a container scheduling system in a micro service environment.Using container scheduling system to schedule and manage the Mall micro services,and tested the performance of the Mall micro service system The test results show that the container scheduling system can not only make the Mall micro service system workload more balanced but also reduce the system response time.