We had about half a billion rows in one table back in 2008, and doing thousands of updates/inserts/deletes per second. Can't remember how much RAN we had, but we sure did have a few spinning disks.
I prefer/dev/null for write heavy workloads that need long term storage. There’s plenty of data in modern physics to suggest that there’s no information loss from going into a black hole, so there shouldn’t be any problems. Put “architected web scale data singularity with bulletproof disaster recovery plan” in your CV. You don’t need to mention the recovery plan is to interview for new jobs when someone starts asking to access the data.