The main idea of this database is that you can request the state of the RPKI at any random point in time and recreate it locally in a form that can be parsed by a relying party client (e.g. Routinator). Data is fetched at a 1-minute interval.
Generating the tar file might take a couple of minutes. The page may seem unresponsive but things are actually happening.
You can also get some data out using some APIs, here are some examples. These might not be stable and still subject to change:
Much like the web 2.0, RPKI Wayback Machine 2.0 is an improvement over 1.0, and will go downhill again at 3.0.