A cloud computing resource allocation strategy based on packet-cluster mapping was proposed. Under the concept of package and cluster, resources can be shared, the task scheduling may become more flexible, and the resource utilization will be raised. Dynamically fusing the multi-objective genetic algorithm with an improved ant algorithm, a cost-optimized cloud computing resource allocation algorithm was provided. In the algorithm, the fast and random global search ability of the genetic algorithm was utilized in the early stage of the task to generate the initial pheromone, and in the later stage of the task, by the ant algorithm, the information among ant colonies was exchanged and a positive feedback mechanism was used to find the optimal solution of the resource allocation. The experimental results show that under the concept of package and cluster, the hybrid scheduling algorithm proposed can significantly reduce the task completion time and task execution cost, effectively reduce the number of cluster nodes used and improve the resource utilization.