This is primarily for scenarios where infrastructure is physically distributed, which could apply in manufacturing, automotive, agriculture, energy, healthcare, telecom, retail, and many other sectors.

For your typical fully-cloud-based scenarios, a lakehouse architecture will meet most of your needs. For higher end, globally distributed services, there are still some advantages to pushing some compute out to the edge. One of the articles I read as I was researching for this article was Microsoft’s Helios Paper, describing the system they have in place for monitoring and debugging issues in their globally distributed CosmosDB service.