“What Goes Around Comes Around... And Around...” by Stonebraker and Pavlo is such an insightful paper. Tons of wisdom!
Curiously, they skipped timeseries databases. I am not sure but maybe timeseries databases are columnar stores, use vectorization etc so not much differentiation from OLAP. This reminds me to take a look at what kind of storage engines these databases use (InfluxDB uses some variant of LSM called Time-structured Merge Tree).
https://db.cs.cmu.edu/papers/2024/whatgoesaround-sigmodrec2024.pdf