Thank you for the details! After all, if I can manage to fit around 6 million points into a single r4.large and get a decent response time when clustering them,

Thank you for the details! After all, if I can manage to fit around 6 million points into a single r4.large and get a decent response time when clustering them, I can't see why I shouldn't try something with it as a secondary data store.