Performance Prediction of Large Data Transfers

Student: Jason Cheung (Stony Brook University, Lawrence Berkeley National Laboratory (LBNL))
Supervisor: Alex Sim (Lawrence Berkeley National Laboratory (LBNL))

Abstract: Scientific facilities around the world transfer terabytes of data to Berkeley Lab’s National Energy Research Scientific Computing Center (NERSC) for processing. These large data transfers can cause congestion on the computer network. To better manage these large transfers, we plan to predict their expected transfer time using machine learning techniques. Through a careful study of traffic logs (Tstat), we find an effective way of utilizing information from recently completed transfers to improve the prediction accuracy by up to 30%.

ACM-SRC Semi-Finalist: no

Poster: PDF
Poster Summary: PDF


Back to Poster Archive Listing