[mwinpc]

(PM at Cockroach here)

There's a little blurb in this blog post about the decision we made at the time https://www.cockroachlabs.com/blog/how-were-building-a-busin... (Note: The blog mentions APL but we've since moved to BSL)

Quote: “The first [of the CCL features] is a fully-distributed, incremental capability for quickly and consistently backing up and restoring large databases using configurable storage sinks (e.g. S3 or GCS). The same functionality, but non-distributed, will be available for free to all users.”

I can't speak to what exactly the thinking was at the time (I wasn't around yet) but as you mentioned, the community told us that the non-distributed functionality (DUMP) wasn't good enough for the scale people wanted to run at. Makes sense, so we made the change.

If I had to take a guess DUMP was performant enough at the time for most of our CockroachDB Core users and their scale, but adoption has picked up since then, as have data sizes.