The biggest feature of Cloud service environment is on-demand delivery, through virtualization technology will be related to build a unified scheduling of resources, and in accordance with the needs of users to provide services, therefore, cloud service has parallel computing, open and on-demand delivery characteristics. For training teaching platform, in the cloud computing environment need to face a variety of users' needs, such as the request of a variety of tasks, the type of experimental tasks are not same, there is a big difference between the device resources, through virtualization technology to achieve standardized management of the resource sharing, cloud resource scheduling can effectively meet the needs of users. Dynamic migration strategy is designed to achieve the management of resources and tasks in the three layers of collaborative resource scheduling mechanism. This research focuses on the task partitioning, resource allocation, resource scheduling strategy, etc., based on the design of the prototype system, the system simulation experiments, is conducted to verify the feasibility and effectiveness of the dynamic migration strategy of training platform in cloud computing environment.