I generally dislike Mongo, but I’ll give an anecdote. Circa 2013 or so I was working for a regional newspaper group. They decided they wanted a weather section

I generally dislike Mongo, but I’ll give an anecdote. Circa 2013 or so I was working for a regional newspaper group. They decided they wanted a weather section. So we subscribed to, iirc, Accuweather. I wrote a script that would pull it down at the update interval (Every 15 minutes I think it was) This was, for the time, for a shoestring outfit with only a few servers, a lot of data. Something like 50k locations, with around 30 “rows” for each location - current conditions plus hour by hour going out a bit more than a day. Loading this into our poor little MySQL server was… slow. So I just setup a small mongo instance, and loaded/read from that. Worked great… we didn’t care about durability so we dodged that whole minefield. Mongo had a way to atomically swap two tables… so the import script would load into a scratch table, then swap that into the one the web server actually read from. Loaded super fast since we didn’t care about durability, and read performance was more than adequate for the very simple queries we ram against it. (To give you a feel for the era, this is when MyIASM was the default table store, and didn’t support fks or transactions)