Page History

...

• Bebo posed the question of whether PingER can identify changes in Internet responsiveness as measured by PingER as a result of the impact of the Covid-19 virus.
• The first attempts looking et PingER statistics and see if there were any noticeable changes in performance attributable to Covid-19 interventions (e.g. shutting down schools,  social distancing, gathering cancellations, lockdowns), were not very successful (see PingER and COVID-19). See https://covid19-interventions.com/ for interventions in European, and some Asian countries.
• Another attempt to correlate the impacts might be to use a tool such as Google Motion Charts to look for correlations between say Covid-19 cases (cases, hospital admissions, deaths etc) and PingER metrics as a function of time, by country, continent, plus population or density of population. The idea is that as Covid-19 cases pass say 100, the country will impose interventions such as school closing,  social distancing etc. which will change the usage patterns of the Internet which may have noticeable impacts on the PingER measurements.
• Unfortunately Google Motion metrics relies on Flash and is no longer supported .
  • A possible alternative is Charta.ca
  • I have therefore put together a Perl script (chart-covid.pl) to take daily PingER metrics for the last 120 days together with up-to-date raw data on the coronavirus from the Johns HopkinsUniversity GitHub site: https://github.com/CSSEGISandData/COVID-19, for various countries with their population, area, population density and cast the data in a format suitable for Charta.ca.
  • There is a web page I put together on how to use Charta.ca at Using Chart.ca.
  • There is an example at https://charteca.blob.core.windows.net/pubs/3Z3qcBXqCT9B3WHz
  • Instead of using Covid-19 confirmed cases, it might be good to use the Unacast cell phone location data to get a measure of how people are social distancing. I have sent them two emails but no response.
    
• Discussion:.
  
  • Les pointed out that many of our targets (and of course the SLAC Measurement Agent) for the US (ESnet and Internet2), Europe (Dante, and national education/research networks etc.), Australia (AARNET) and parts of Asia (TEIN) are on well-provisioned, lightly loaded networks so it is the transit networks are not expected to see big changes.
  • Umar pointed out that the closing of schools, people staying at home should increase the traffic (e.g. between students and profs teaching remotely, people surfing the web watching videos etc) on CDNs (Content Deliverable Networks), and much of this traffic is on public networks. Thus one might see big changes for traffic transitting public networks. 
  • Since PingER is not monitoring homes and content deliverers, PingER would not expect to see big changes in responsiveness for most of our targets in the above regions.
  • On the other hand, African networks are less well developed and PingER may see more of an effect for such targets.

...

...