In scope of this research the following protocols have been investigated for its efficiency on 10Gbps links: UDT, RBUDP, MTP and RWTP. Caused of it another transport protocols have been developed and successfully used for e.g. In nowadays networks, TCP fits not all communication needs that society has. The contemporary common TCP needs significant improvement since it was developed as general-purpose transport protocol and firstly introduced four decades ago. This work is dedicated to comparison of open source as well as proprietary transport protocols for high-speed data transmission via IP networks. We show that for XFS and Lustre with direct I/O, this method identifies configurations achieving 97% of the peak transfer rate while probing only 12% of the parameter space. We propose a new method to efficiently identify the optimal joint file I/O and network transport parameters using a small number of measurements. Furthermore, large variations in the measured rates necessitate repeated measurements to ensure confidence in inferences based on them. Our results indicate that large buffer sizes and many parallel flows do not always guarantee high transfer rates. We present extensive measurements of disk-to-disk file transfers using Lustre and XFS filesystems mounted on multi-core servers over a suite of 10 Gbps emulated connections with 0–366 ms round trip times.
It remains a challenge to achieve peak rates for such transfers due to the complexities of file I/O, host, and network transport subsystems, and equally importantly, their interactions. File transfers over dedicated connections, supported by large parallel filesystems, have become increasingly important in high-performance computing and big data workflows.
0 kommentar(er)
