It feels very much like they started out wanting to build the AWS version of Spanner/Cockroach/Yugabyte/Foundation/Tidb but its characteristics were diluted by committee to something that nobody wants or needs.

There is no point having an opaque-partitioning distributed database that sits at Repeatable Read.

Meanwhile AWS still suffers from want of a real alternative to GCP Spanner (still the most compelling reason to choose GCP over AWS).

Curious to hear any challenges to my perspective.