The point where people start to use UMH is, if they have already done some PoCs, but then realize that their approach with Node-RED etc is not scalable, and they start to loose overview. Also when it gets hard to involve more people to it, and think about maintenance (more than just install and forget). Typically, these discussions start when someone experiences a botched update (e.g., Node-RED or OS) killing an entire application, or similar situations (InfluxDB loosing data, etc.) In those cases, leave everything as it is, but additionally install the UMH. Then you could start routing data to it, and see if the centralized management, guaranteed software lifecycle etc makes sense for you. Then you could slowly migrate your old use-cases to the new infrastructure. Also, the everything ignition approach does not scale, but using ignition as "magic UNS" does. Basically, you can keep all your software and use-cases, but simply get less maintenance burden.