Joint Optimization of Caching and Routing Strategies in Content Delivery Networks: A Big Data Case

Content delivery networks (CDNs) have been proposed to improve the performance of large-scale content delivery in communication networks, in which content files are dynamically cached in CDN nodes that are close to end-users so as to decrease the transmission latency and traffic redundancy. We investigate a real CDN that is currently utilized by a large social network company in China. We note that there are two important issues that are not well considered in the existing literature. The first is that end-users can usually access multiple CDN nodes with high quality of experience. The second is that the data service price may vary greatly for different regions, which can highly influence the cost of CDNs for large-scale applications. We reconsider the optimal caching and routing problem in CDNs by jointly considering these two practical issues, and propose a joint caching and routing strategy by using the alternating optimization technique. Simulation results show that the proposed algorithm outperforms the current CDN strategy, in which most popular files are cached and end-users are directed to the nearest CDN nodes, by 30% and 12% in terms of latency and data service cost, respectively.