Most orgs, including mine, also have the experience of a cassandra cluster simply melting down. That was more common ten years ago, but it is pretty much chaos

Most orgs, including mine, also have the experience of a cassandra cluster simply melting down. That was more common ten years ago, but it is pretty much chaos when it happens. If the non-SQL distributed databases, I found FoundationDB to be by far the most robust and zero-overhead. However, at least through version 6, it had absolutely terrible, service-destroying behavior if any of the nodes became low on storage, dropping transaction levels through the floor and since transactions are used for recovery and cluster expansion... Since it does a very nice job auto-sharding, if you're not paying attention this tends to happen to multiple nodes around the same time. I wish it would get more attention though. It is a really amazing tool.