A Multi-Perspective Study of Internet Performance during the COVID-19 Outbreak

Abstract: The rapid spread of the novel corona virus, SARS-CoV-2, has prompted an unprecedented response from governments across the world. A third of the world population have been placed in varying degrees of lockdown, and the Internet has become the primary medium for conducting most businesses and schooling activities. This paper aims to provide a multi-prospective account of Internet performance during the first wave of the pandemic. We investigate the performance of the Internet control plane and data plane from a number of globally spread vantage points. We also look closer at two case studies. First, we look at growth in video traffic during the pandemic, using traffic logs from a global video conferencing provider. Second, we leverage a country-wide deployment of measurement probes to assess the performance of mobile networks during the outbreak. We find that the lockdown has visibly impacted almost all aspects of Internet performance. Access networks have experienced an increase in peak and off-peak end to end latency. Mobile networks exhibit significant changes in download speed, while certain types of video traffic has increased by an order of magnitude. Despite these changes, the Internet seems to have coped reasonably well with the lockdown traffic.