KoutenDB v0.9.0 adds benchmarked related-data reads against PostgreSQL

Developer puffball1567 has released KoutenDB v0.9.0, a ring-oriented NoSQL document and vector database written in Nim, as a technical preview on GitHub. The release focuses on testing whether the database's locality model can handle realistic, persistent related-data reads rather than remaining a theoretical design concept. Version 0.9.0 introduces a 'subring' query mechanism that lets callers retrieve multiple bounded, related collections — such as a user's profile, orders, and notifications — in a single read operation with independent limits per collection. A benchmark comparing KoutenDB against PostgreSQL on 10,000 users and 1,050,000 logical records showed read latencies of roughly 197 microseconds for KoutenDB, 515 microseconds for six separate PostgreSQL indexed queries, and 236 microseconds for a PostgreSQL JSON aggregate query, all on a single test machine. The release also includes performance improvements such as stream caching for disk-backed reads, projection-state reuse across subrings, and chunked import support via a new batch-size flag.