What if we used contact tracking instead of contact tracing? How would it scale?
For my Master's thesis, I needed a system to benchmark and run as an example, to check whether my configuration management tool worked correctly. Because of this, I decided to explore whether contact tracing would be a better solution - whether GPS data from mobile devices of infected individuals could be aggregated responsibly to generate heatmaps, so even those without applications would know which parts of the city to avoid, as well as something that could aid in the research on the spread of the virus.
What I did
I created a Ruby on Rails application that in turn integrated with a PostGIS database. This application would then respond to REST requests and aggregate the received data for further processing into heatmaps, so that the individual device positions couldn't be discovered, but the overall picture of things would still be available.
What I learnt
Surprisingly, this sort of system worked rather well and didn't have that many issues scaling, even when running on just a few comparatively affordable VPS nodes. Of course, in general, the runtime will be slower than something like JDK or CLR, but it's still impressive, regardless.