Abstract:CoreOS is a new containerized cluster server operating system based on Docker. It is developing rapidly and has been supported by mainstream cloud service providers such as OpenStack, kubernetes, Salesforce and Ebay. The load in the cloud environment is dynamic, therefore the resource requirement is dynamic, which poses a challenge to the efficient utilization of cluster resources. The strategy of static pre-allocation of peak resources brings a huge waste of cloud resources, and in the meantime idle calculation wastes a lot of energy consumption. In this paper, load-integrated cluster scheduling system (LICSS) is proposed to monitor the load distribution of the cluster in real time. In order to release idle resources in time to reduce energy consumption in the scheduling process, the nodes are allocated using the compact scheduling strategy and the load is dynamically integrated by the task migration technique. The LICSS system implements the node load metric, task metric, load integration algorithm, and calculates the node adaptive load threshold. Experiments show that the LICSS system can effectively integrate the load according to the dynamic changes of the cluster load in different time periods, and can improve the average resource utilization rate by 12.2%, and reduce the cluster energy consumption by triggering the dormancy of the redundant nodes during the low load period.