SURFnet is logging its network failures. We propose to analyze those failures to better understand what their impact was, and to investigate if there are certain correlations between events (like maintenance) and failures, such that SURFnet can better protect against similar events in the future.

The outcome of our study will clearly depend on the data that will be made available. It is our hope that the data is sufficiently rich to make clear whether the network has become more robust (i.e., has been less affected by failures) over the past years, and which protection mechanisms have been (most) effective. Our study will also scrutinize the way of logging failures, and possibly give recommendations to ascertain that, for the future, enough information is present in those files to be able to learn from past failures.

If possible, we would also like to see (possibly with the help of SURFnet) if more data can be obtained, for example from RIPE NCC and Agentschap Telecom.