A method for enhancing end-to-end transfer efficiency via performance tuning factors on dedicated circuit networks with a public cloud platform

Abstract

There has been a great deal of recent research interest regarding the storage and utilization of big data via remote cloud platforms. The efficiency of large data transfers from remote cloud platforms is a critical issue, and dedicated networks areused for data transfer. To resolve the data transfer efficiency issue, it is necessary to tune the L2-related performance items in the transport equipment and to regulate performance factors, such as window size, between IP layer servers, during the configurationof an end-to-end research network. It is also necessary to tune factor values related to performance items in transport equipment. System level kernel parameter tuning is also needed and results in improved throughput. Here, we measure throughputaccording to L2-related tuning factors and IP levels, including kernel parameter turning, and present an analysis of the measurement results. The experimental results show that end-to-end servers with tuned factors, in addition to system level kernelparameter tuning, can effectively utilize the available bandwidth.