Docker container has been used in cloud computing at a rapid rate in the past 2 years, and Docker container resource scheduling problem has gradually become a research hot issue. It is NP-complete as the optimization criteria is to minimize the overall processing time of all the tasks. Nevertheless, minimization of makespan does not equate to customers’ satisfaction. Aiming at the performance optimization of Docker container resource scheduling, the authors propose a multi-objective container scheduling algorithm, namely Multiopt. The algorithm considers five key factors: CPU usage of every node, memory usage of every node, the time consumption transmitting images on the network, the association between containers and nodes, the clustering of containers, which affect the performance of applications in containers. To select the most suitable node to deploy containers needed to be allocated in the scheduling process, the authors define a metric method for every key factor and establish a scoring function for each one and then combine them into a composite function. The experimental results show that compared with the other three well-known algorithms: Spread, Binpack, and Random, Multiopt increases the maximum TPS by 7% and reduces the average response time per request by 7.5% while consuming roughly same allocation time.