Intelligent Action Performed Load Balancing Decision Made in Cloud Datacenter Based on Improved DQN Algorithm

Due to dynamic changes of cloud state and increases user demand, load of datacenter fluctuated at regularly that shows load balancing problem. It is a challenging issue to take an appropriate action by datacenter controller to reduce the processing time of all incoming task with allocating the best resources in minimum time period. Therefore, an effective task scheduling is required to balance the load in datacenter. This paper proposed an Improved Deep Q-Network (I-DQN) task scheduling algorithm to balance the load. In this algorithm agent take a suitable action that minimize the makespan time. Simulation is done by using Google Colab with Tensorflow show the effectiveness of proposed scheduling algorithm. From the experiment we show our proposed algorithm is better success rate with reduce makespan time, waiting time and throughput as compare to existing DQN algorithm.