The fact that HTTP is the most used application protocol in the Internet is well known. However, statements like this are seldom supported with clear evidence. At the Design and Analysis of Communications Systems (DACS) group, of the University of Twente, we have created ReFlow: a public website containing statistics about general Internet traffic. The main goal of ReFlow is to maintain historical traffic data, allowing for comparisons and analysis of trends on Internet traffic. This way, basic facts on the Internet can be supported with evidence.
ReFlow was conceptualised in 2012 and since has been in constant development. The first steps were taken by Martijn Hoogesteger, as part of his Bachelor project at the university. During the TERENA Networking Conference in Maastricht ( TNC 2013 ), we presented the project. It was very well received and supported by the community. Over the last year we worked hard to release a stable version of ReFlow. This was accomplished and ReFlow was once again part of the TNC program (at TNC 2014 in Dublin), Please see below the poster we presented. Again, we received positive and constructive feedback, for which we are very grateful!
Statistics reported on the ReFlow website are calculated over traffic aggregates from several sources. We use NetFlow-style traffic measurements, widely available in network devices nowadays. Currently, we receive traffic reports from many collaborators around the globe: RNP from Brazil, DeiC from Denmark, CESNET from Czech Republic, University of Twente from the Netherlands and GÉANT. We are looking to extend our collaborators list and gather traffic information from as many sources as possible. If you would like to send us data, please contact us .
Notice that ReFlow does not use sensitive traffic data. For example, calculated statistics do not make use of individual IP information or IP connections. In fact, ReFlow is implemented such that we do not even have access to the NetFlow data captured at the operator’s side. We provide open-source scripts that read NetFlow captures, process statistics, and send us a ready-to-use report file. This report file only contains the resulting statistics.
Statistics on Internet traffic are calculated on a weekly basis. Among others, these include: distribution of traffic per port, traffic per IP protocol, and CDF (cumulative density function) plots of flow duration, flow throughput.
All statistics we compute over time are made publicly available at: stats.simpleweb.org . So if you want to follow our project, check our website regularly! Who knows, perhaps we will see some interesting things happening on the Internet in the next few years. Will HTTPS slowly take over HTTP? How much IPv6 tunnelling traffic is present in general? Does traffic look different in Brazil compared to Europe? These are all questions that ReFlow could answer.