I liked his talks a lot, ended up using a rudimentary event sourcing system in my own hobby project.
It's not super big, but there are over 30 million events in there. It's all just running on mariaDB so no fancy software in there, there's this `stored_events` table and it just keeps on trucking. Software running fine :)
It's quite nice to keep history. I never made tables for actual payments, but I did store paypal events in this table.
Only recently did I realize this data is actually useful for me as support for my users so I made a UI loosely based on this query, and it just works
SELECT somestuff FROM stored_events WHERE event_type='paypalIPNSuccessfulEvent';
I made this event easily over 5 years ago, today the data is useful. Thank you event sourcing!
Edit: Oh and I shouldn't forget that in between this time I was constantly thinking about how big of an architectural mistake this was. I was just keeping a bunch of data, not using it, using some abstraction sold to me by some guy on YouTube...
Well, it has its advantages, but the biggest issues lied with how I thought the table had large performance issues. Turns out it doesn't really have performance issues, I just didn't know how to use indices :|
The aforementioned SELECT statement runs in milliseconds, where there are a couple thousand of that specific event out of 30m+