The use of content delivery networks is emerging with the increasing population with access to the Internet. Content Delivery Networks bring content closer to the end-user lowering operational expenses and increasing end-user experience. End users consume content from the cache servers using redirection. Optimising the load distributed to these cache servers is crucial to maintain a sustainable network. In this article, we created and evaluated a new routing method with the help of artificial intelligence exploiting Software-defined networking. The new routing methods aim to distribute the load to the delivery servers optimally and maximises utilisation of the network, avoiding bottlenecks along the way. The proposed approach distinguishes and routes the flows in the network on the session layer rather than just following standard routing rules based on prefix matching. Concluding with the evaluation of achieved results and the gained performance increase in this article.

In today’s age of digital media and social networks, delivering content to a massive scale of recipients became one of the main challenges. Load Balancers and Content Delivery Networks are used to distribute content to multiple locations, which increase the scalability of services and decrease the load on content origins. Both technologies rely on redirections. Redirections have not received a significant amount of attention in the recent years; however, they do impose some limitations. In this article, we propose a transparent redirection mechanism, which exploits the versatility of Software Defined Networking. The redirection method is achieved by handing off existing TCP sessions without any required modifications to the endpoints. This article demonstrates how the proposed redirection mechanism can be adopted in Content Delivery Networks and Load Balancing scenarios. The performance of the solution is thoroughly tested and compared to existing legacy solutions.