Hey HN, I built Bingsan as an open-source Apache Iceberg REST Catalog.

Why? Existing options (Nessie, Polaris) are Java-based. I wanted something lightweight that runs alongside PostgreSQL without a JVM.

Stack:

- Go + Fiber framework

- PostgreSQL with pgx/v5 connection pooling

- Prometheus metrics built-in

Performance (Xeon 8581C): - ~24k req/s throughput

- 44μs p50 latency

- Linear scaling to 200+ concurrent connections

Multi-node:

- Run multiple instances behind a load balancer

- PostgreSQL advisory locks for leader election

- Background tasks coordinate automatically

Quick start:

  docker run -p 8181:8181 ghcr.io/teampaprika/bingsan:latest

Works with Spark, Trino, PyIceberg. Supports S3/GCS, OAuth2, API keys, multi-table transactions, server-side scan planning.

GitHub: https://github.com/teamPaprika/bingsan

Docs: https://teampaprika.github.io/bingsan/en/

Happy to answer questions!