The energy-related issues in distributed systems that may be energy conservation or energy utilization have turned out to be a critical one. Researchers worked for this energy issue and most of them used Dynamic Voltage and Frequency Scaling (DVFS) as a power management technique where less voltage supply is allowed due to a reduction of the clock frequency of processors. The cloud environment has multiple physical hosts, and each host has several numbers of virtual machines (VMs). All online tasks or service requests are scheduled to different VMs. In this paper, an energy-optimized allocation algorithm is proposed where DVFS technique is used for virtual machines. The fundamental idea behind this is to make a compromise balance in between energy consumption and the set up time of different modes of hosts or VMs. Here, the system model that includes different sub-system models is explained formally and the implementation of algorithms in homogeneous as well as heterogeneous environment is evaluated.