I agree with the article that most data sets comfortably fit into a single traditional DB system. But I don’t think that implies that big data is dead. To me big data is about storing data in a columnar storage format with a weak schema, and using a query system based on partitioning and predicate push down instead of indexes. This allows the data to be used in an ad-hoc way by data science or other engineers to answer questions you did not have when you designed the system. Most setups would be relatively small, but could be made to scale relatively well using this architecture.