It seems that newcomers to distributed systems are often confused by the various definitions of the C-word. I want to go over them to draw a distinction that is often overlooked, without getting into the tired old arguments around eventual consistency. Specifically, I want to expand on the distinction between properties of a data store and the properties of applications built on top.
The C in CAP, the C in ACID and the C People Care About Are Not the Same.
First time I read such a clear write-up of the consistency differences.