[names_are_hard]

I'm not GP so I don't know what they meant, but a key upside to complex logic in the application layer vs the database is that the application layer is often much easier to scale out than the db. Where I work, if I run out of memory in the app I just change a configuration variable and k8s gives me more instances instantly. But if the database is memory constrained and I'm already on the biggest server available to me...I need to re-shard or take some other more sophisticated approach (tuning, replication, other stuff I don't know about).

Of course each scenario is different, YMMV, and as always "it depends".

[pdhborges]

> key upside to complex logic in the application layer vs the database is that the application layer is often much easier to scale out than the db

I think the point of the GP is that all of these application instances are still connected to the DB, doing sub optimal data fetches taxing the database in multiples.