Logistics of High Throughput Data Transfer in High Energy Physics

High Energy physics analysis requires a large amount of data, which is increasing at an ever faster pace. Often this data is not stored at the computing resource where it is meant to be processed especially when there is a need for GPU's due to usage of machine learning, leading to a need for fast, robust and reliable methods of data transfer. Existing methods range from scp which has the advantage of being simple to configure and transfer, to GLOBUS which allows for optimization of system resources. This poster compares the various methods for various use cases.