Importantly, in analytics workloads, it is very important to know roughly how many writes aren't making it. Otherwise your analytics system sucks.

Interesting. How would one know that?

Good question. You'd need some accurate-enough data source telling you about failed writes. Which eventually comes back around to needing a consistent database and indications of client disconnects.

With a huge amount of data (as I've heard analytics is), could you take a sampling approach where you log every n transaction and only check those against the DB?