All you need is PostgreSQL
PostgreSQL alone can serve as a comprehensive backend for many applications, replacing tools like Redis, Kafka, and Elasticsearch. Its built-in features such as NOTIFY/LISTEN, full-text search, and partitioning reduce architectural complexity. The author recommends favoring PostgreSQL's integrated capabilities before adding external dependencies.
Background
- This blog post argues that many modern tech stacks overcomplicate things by using a separate database, cache, message queue, job scheduler, and vector database — when PostgreSQL alone can handle all those roles.
- PostgreSQL (often called Postgres) is a free, open-source relational database that has been steadily adding features like full-text search, JSON storage, pub/sub messaging, and vector similarity search (via the pgvector extension).
- The piece is part of a long-running debate in software engineering between "use the best tool for each job" and "keep your stack simple with fewer moving parts." The author leans heavily into the latter, advocating for a "Postgres-centric" architecture.
- Key context: Postgres is widely respected but historically wasn't seen as a specialist in things like caching (where Redis dominates) or search (where Elasticsearch dominates). Recent extensions have blurred those lines, making this argument increasingly plausible.