As someone who has rolled their eyes many times at people calling dozens or hundreds of gigabytes of relatively simple and well-structured data big data I wholeheartedly agree.
I think the only thing that’s big is the prices cloud providers charge to some of these companies for managed solutions, sometimes even while making things like upgrades of databases harder than when self-hosting. So I think big data has mostly become a marketing term for relatively inexperienced people/companies.
But I am very happy that a lot of the trends mentioned in the article are dying again. But I don’t see that trend purely negatively. I think there have been a lot of learnings from it. Not sure it can be directly attributed to this, but traditional systems put some effort into making multi-machine setups easier, and now cover some use cases that were poorly covered before that trend. Be it multi-machine setups, simplifying things like HA setups, sharding, logical replication, etc., or even JSON instead of just things like hstore, since MongoDB was mentioned. I even think that the awareness of indices and performance characteristics has risen.