A Cost-Based Distributed Algorithm for Load Balancing in Content Delivery Network

Load balancing plays an important role in improving scalability and stability in Content Delivery Networks (CDNs) to meet the increasing demand on bandwidth. This paper proposed a modified algorithm that takes into account the equilibrium between load balancing and redirection proximity. We extended a fluid queue model which is adopted in the existing literatures to the overall CDN system. In the system, scheduler selects proper replica server for each redistributed request by exploiting load differences between them. Furthermore, through limiting the migration distance for each request, the total costs mainly associated with delay are also effectively optimized. The simulation result indicates that the proposed algorithm can efficiently reduce redirection cost compared to Control-Law Balancing (CLB) algorithm at the expense of a bit of performance sacrifice of queue balancing. Besides, we found that the proposed mechanism has more benefit on queue balancing than CLB algorithm as well, when selecting an appropriate distance threshold.