It's true that technically we already have all the data we need. However, the problem I always encounter is that this data is very siloed and therefore hard to combine. The value of a UNS/UMH is the ability to combine L2 process data from SCADA with L3 data from an MES. This allows me to assign process values, such as temperature, to a particular batch (as opposed to only a timestamp). In my past data engineering job we didn't have a UNS. We were manually building data pipelines to bring this data together in another system in the cloud. Figuring out how all of this worked, and then programming the point-to-point connections was extremely time-consuming - and therefore super expensive. In fact, 80% of our advanced analytics projects were really just data plumbing. If you can make data integration just 20% of your project duration instead of 80%, you can deliver more projects faster. Data engineering labor is expensive: imagine the impact this has on the bottom line of a company.