Real-time genetic optimization of large file transfers

Transfer configurations play a significant role in achieved throughput for file transfers in high-speed networks. However, finding an optimal setting for a given transfer task is an intractable, non-linear problem. Existing solutions thus rely on offline models to configure only a subset of parameters, yielding suboptimal performance when network conditions deviate from what is observed in historical data. In this paper, we apply genetic algorithms to discover optimal configurations for file transfer settings in real-time. Experimental results show that the genetic algorithm yields up-to 33% higher transfer throughput compared to the state-of-the-art solutions by finding near-optimal transfer settings with minimal overhead.