An Improved Energy Saving Strategy for SDN-based Data Center

With the rapid development of IoT (Internet of Things) technology, the energy consumption problem of IoT data center is getting worse. Based on the fact above, a decision tree based energy optimization strategy CARTS (Classification And Regression Tree based on SDN) is proposed in this paper. The strategy which is based on the Gini index constructs a decision tree from the users' requests' set. Pruning is performed via calculating the penalty factor of the decision tree. Then an optimal decision tree is obtained. Based on this optimal decision tree, CARTS obtains the relationship between users' requests and responses, thereby access efficiency is improved. Efficient access leads to total storage reduction. Energy saving is gained. In addition, the SDN is adopted as the data center's network architecture. The energy saving strategy is easy to be deployed in the SDN controller. Finally, the simulation shows that the energy optimization strategy CARTS proposed in this paper is more energy efficient than the traditional typical energy optimization strategies, and the energy saving ratio is up to 11%.