The original title is "The blueprint: an HTML-to-PDF API on Postgres, Chromium, and not much else"

Short summary

A solo founder details the architecture of DocPenny, an HTML-to-PDF API built on Postgres, Chromium, and S3 with no Redis or Kubernetes. Key decisions include using pg-boss for job queuing on existing Postgres, streaming PDFs in 1MB chunks to keep memory fixed at ~20MB per document, and encrypting sensitive payloads with AES-256-GCM. The system handles ~5,900 documents/s ingestion and ~63 renders/s on a €45/mo server.

• •pg-boss on Postgres eliminates need for a second stateful service like Redis
• •Streaming PDFs from Chromium to S3 in 1MB chunks keeps memory fixed regardless of document size
• •Queue absorbs 90x ingestion-to-render gap; never couple accept rate to render rate

Generated with AI, which can make mistakes.