On Multi-CDN Delivery Costs Optimization Problem

This paper studies the problem of minimizing delivery costs in multi-CDN streaming systems. As inputs, this problem receives information about minimum edge traffic volume commitments and edge traffic rates defined in CDN contracts. As input, the problem also receives historical data about CDN traffic volume statistics and predicted volume trends for the remainder of a billing period. As output, it computes load allocation for all CDNs for the remainder of the billing period, minimizing total CDN delivery costs. We show how this problem can be formalized and analytically reduced to a superposition of linear programming and combinatorial search problems. As such, this problem is tractable and practically solvable by existing methods. The paper also discusses the implementation of the proposed methods by multi-CDN systems using the HLS/DASH content steering mechanism.