Application of Deep Learning in Dynamic Link-Level Virtualization of Cloud Networks Through the Learning Process

In this paper, we first present the architecture of small cloud data center network (DCN). We then simulate strategies of link aggregation between two virtual switches in 8 and 16 server DCNs. We use an emulation tool to include the impact of software defined network (SDN) controller. Focusing on links, we use link aggregation control protocol (LACP) technique that results in simulating traffic analysis such as traffic forwarded and received bits per second, average throughput on switches. We then use an artificial intelligence (AI) tool to apply the results of the simulation to a deep learning (DL) paradigm. Deep learning implements training and applies the percentage traffic split on links and the time that links need to be aggregated to achieve optimal link-level virtualization. Deep learning automatically detected performance degradation and bandwidth requirements and implement proper configuration changes needed in the network. This article then presents the application of the deep learning incorporated into the virtualization technique when using the link aggregation control protocol (LACP) on links of cloud networks. The process is carried out through self-learning and training processes that help in building fault-tolerant and low latency networks.