Not saying that this technique doesn't have any use, but the example scenario is pretty weak. Why not just send s3 link to the analysts
s3://robotaxi-inc/daily-ride-data/*.parquet
Or just the query: SELECT pickup_at, dropoff_at, trip_distance, total_amount
FROM 's3://robotaxi-inc/daily-ride-data/*.parquet'
WHERE fare_amount > 100 AND trip_distance < 10
does the intermediate database and view really provide much value?Thinking more, I feel this boils down to ownership. Does it make sense for you to own this abstraction layer? Or does it make sense to shift the ownership towards the receiving end, and just provide the raw data.
To me maintaining this sort of views to data sounds more like responsibility of data analysts that random devs. They are the experts in wrangling data afterall. But of course there is no single right or wrong answer either