Construction of Virtualization Adaptive Algorithm for Virtual Machine Environment

In today's virtual machine environment, dynamic resource requirements are often unbalanced, which may lead to inefficient resource utilization. Therefore, this paper proposes an algorithm to optimize dynamic resource allocation. This paper first constructs a load prediction model based on long short-term memory (LSTM) to estimate the resource requirements of virtual machines. Secondly, this paper also introduces a deep Q network (DQN) strategy to ensure that the system can operate efficiently in a complex and changing environment. Finally, this paper proposes a priority-driven resource scheduling mechanism to ensure the balance of system performance. In the experimental conclusion, the resource utilization of LSTM-DQN is maintained between 60% and 90%. Under high load conditions, the response time of LSTM-DQN is 250 milliseconds. In addition, the load fluctuation standard deviation of LSTM-DQN is 2.0, which is much lower than the static allocation of 0 and DQN's 7.8, indicating that it has better load balancing ability. The experiment verifies the effectiveness of LSTM-DQN in the virtual machine environment.