[cheez]

Ah yes, embarrassing not to remember basic DB architecture :-)

Do you have an example of a workload that benefits from this structure?

[kyllo]

Yes, it's geared toward reporting / analytics / data science workloads where the most common read operation is calculating aggregate metrics like count/min/mean/max and histograms. Statistics stuff.

[cheez]

Thank you very much for your patience :-)

[nuclearnice1]

Taking stats on a collection of time-ordered set of sensor observations.