a year in lockdown: how the waves of COVID-19 impact internet traffic (PDF)
if you’re in the internets biz, this is well worth sitting down to read.
summary
a group of european researchers look at traffic data from a variety of sources in the EU. an IXP, a university network, a large EU ISP, and a mobile operator. they have the benefit of a couple of years worth of flow and traffic records for comparison. analysis is broken down a number of ways and they do a nice job of correlating behaviors across the networks that they have access to.
personal notes/reactions
given their sampling locations, they’re likely missing out on some percentage of the most interesting developments and i suspect they significantly undercount the actual increases in network traffic.
i strongly suspect that they’re missing the biggest traffic spikes coming from private peering with content and cloud providers. though to be fair the central european ISP application traffic mix seems to capture the most interesting elements and really is the most interesting aspect of the analysis. this is pretty much the sole sample they have for visibility into these traffic destinations.
my only real quibble here is that they don’t seem to provide any read into the traffic destined to/from the big cloud/social media providers. they aren’t particularly clear as to how the traffic data is collected in the most interesting network, the ISP.
still, the general trends are broadly useful. it would be great if we could get complementary and x-ref’d data from the major traffic sources in the world to these networks.
meta
- location: Minneapolis, MN
- weather: ☀️(Sunny) +72°F