Optimization of Monitoring and Scheduling Strategy for Distributed Docker Cluster

Swarm is an official native cluster management tool for Docker, which integrates multiple Docker nodes into a virtual single host and provides container distribution and scheduling management capabilities. Analyzing the overall architecture of Swarm, Spread as the default scheduling strategy encounters issues such as resource fragmentation and load balancing, a statically balanced improved container scheduling algorithm is proposed. A Swarm cluster monitoring system is constructed to monitor the performance of containers and node resources in real-time. Based on the resource request amount of containers to be scheduled, the weighted average of the resource utilization rate and load balance of candidate nodes is calculated as the scheduling weight. Experimental results show that compared with Spread strategy, the proposed method can reduce resource fragmentation on nodes, achieve better cluster balance, and enhance the overall performance of the cluster.