PerfSONAR data representation and analysis

• Main Author: Goh, Ling Yong
• Other Authors: Lee Bu Sung
• Format: Final Year Project
• Language: English
• Published: 2017
• Subjects: DRNTU::Engineering::Computer science and engineering
• Online Access: http://hdl.handle.net/10356/69992
• Institution: Nanyang Technological University

The internet is now a basic service similar to water and electricity. At the very core of the internet is the “computer network” consisting of routers and switchers. Billions of dollars are potentially lost if the network is down for a day. For just Google company, almost $100 million can be lost from advertisements if it stops operating for a day [5]. Computer network is an important infrastructure for commerce, defense, social networks, media and etc. Thus, it is essential to monitor and act proactively to ensure the network is in operation. Rather than just looking at current traffic, it is also to predict the network traffic. So one of the way is to use an automated prediction/forecasting algorithm to predict how the network should look like. Holt-Winters is chosen to be the algorithm with its capability to perform exponential smoothing. In this project, I have worked with the network engineer of Singapore advanced research and education network(SingAREN) to develop a tool to predict the network traffic. The tool gets information from the API and displayed it on a table and graph. Similar information is grouped together. Network engineers can use the provided graph implementation and prediction to make informed choices based on the data analyzed and represented. Error information are shown to network engineers with three statuses (Critical, Warning, Minor). As data from the network can be huge and sieving through the data can be a huge hassle, with the new information in hand, they can create faster and possibly better solutions for the network through the help of the project.