Observability in DevOps - What Should I Monitor?

There are lots of things to monitor, right? They're logging everything and then what should you think of one day? What does the a lot's help look like in ML observability and how can teams navigate it? I would say input data drift is a really powerful thing to monitor. And at YLogs, we monitor for the number of nulls, cardinality changes,. All of these things that would indicate there's something going on with my data.