SeederTrading: Trading Swarm Capacity for Improving Content Distribution

A surplus seeding capacity is often observed in large swarms with many seeders while small swarms with few seeders suffer from low download performance. In this paper, we propose a multi-swarm collaboration approach, called Seed erTrading to improve overall content distribution performance by exploiting non-uniform distribution of seeders over swarms. In a nutshell, we enable the over-seeded swarms to share their surplus seeding capacity with the under-seeded swarms. To realize the SeederTrading, we resolve several issues including estimation of swarm capacity, selection of swarms to be matched, and seeder trading across the selected swarms. Our extensive measurement shows that 12.9% of under-seeded swarms can achieve their maximum download throughput by utilizing the surplus seeding capacity. Simulation results show that the under-seeded swarms can improve the download performance by over 20% while retaining the content distribution performance of over-seeded swarms. We also show that the trading seeding capacity across swarms can be done within 60 seconds.