The present application discloses a method and apparatus for elastically scaling a virtual machine cluster. A specific implementation of the method includes: calculating a first amount of residual resources of virtual machines in a virtual machine cluster during service deployment; acquiring an amount of resources demanded by a service to be deployed; acquiring a configuration level of the virtual machines in the virtual machine cluster; determining a virtual machine having a predetermined configuration level as a target virtual machine based on whether the first amount of residual resources satisfies the amount of resources demanded; and deploying the service to be deployed to the target virtual machine. This implementation can implement service deployment in a virtual machine cluster including virtual machines having different configuration levels and also can implement the deployment of different services on a single virtual machine, thereby reducing the costs of service deployment and improving the resource utilization rate of the virtual machine cluster.