Network optimisation using DTN
Boey, Bryan Shao En
Date of Issue2018-11-07
School of Computer Science and Engineering
Large datasets are transferred as a daily routine every day, especially for big companies. The endpoints for the data transfer would vary from national to between countries, which would mean high latency, causing a long fat pipe problem. The objective of this project is to create a purpose-built computer system, to achieve high performance in the transferring of these large datasets, which is dedicated to wide area data transfer to solve the long fat pipe problem. This computer system is called a Data Transfer Node. The scope of this project is to optimize the throughput of the Data Transfer Node. The project consists of 3 parts: Network simulation using ns-3, tuning of hardware in a closed network and last but not least, live testing of the Data transfer node. Results from the network simulation of the ns-3 will assist in the tuning of the network while benchmarking and tuning of the file system will assist in the tuning of the hardware. In network tuning, the TCP settings are tuned in their send and receive buffer size, which takes into considering of the latency and bandwidth of the connection. For hardware tuning, the benchmarking of different file system settings shows the best setting to be used to achieve a higher throughput from the DTN. This project outcome shows the difference between an untuned system compared to a properly tuned system. It also shows how the DTN is limited by its hardware depending on the performance of the hardware and whether it is properly tuned. Recommendations for future work would be to increase the security by implementing security certificates and the automation of file transfer between the host and destination.
DRNTU::Engineering::Computer science and engineering
Final Year Project (FYP)
Nanyang Technological University