Overview
You add Redis for caching, Elasticsearch for search, and Sidekiq for jobs, and soon your app depends on many datastores. Each one needs its own deployment, backup, monitoring, and incident response, so operational cost climbs fast.
postgresisenough.dev makes a simple point: check whether Postgres already does the job before you add another database. This post sums up those per-use-case alternatives.
Alternatives by use case
Before you reach for another system, see whether a Postgres feature already covers the need.
| You need | You reach for | Postgres has |
|---|---|---|
| Caching | Redis, Memcached | UNLOGGED tables, materialized views |
| Job queues | Redis + Sidekiq, RabbitMQ | SKIP LOCKED, pgmq, pgflow |
| Full-text search | Elasticsearch, Algolia | tsvector, pg_trgm, ParadeDB |
| Document store | MongoDB, CouchDB | JSONB, FerretDB |
| Vector search / AI | Pinecone, Weaviate | pgvector, pgvectorscale |
| Time-series data | InfluxDB, TimescaleDB | TimescaleDB, pg_partman |
| Analytics / OLAP | Snowflake, BigQuery | pg_analytics, DuckDB integration |
| Graph database | Neo4j, Neptune | Apache AGE, recursive CTEs |
| Geospatial | Specialized GIS systems | PostGIS |
What each feature does
Here is a short note on each Postgres feature in the table.
Caching
UNLOGGED tables skip the WAL, so writes run fast; a crash wipes them, which suits a volatile cache. Materialized views store heavy precomputed results and serve them fast until you REFRESH.
Job queues
SELECT ... FOR UPDATE SKIP LOCKED grabs rows and skips locked ones, so many workers pull jobs without fighting. pgmq adds a queue, and pgflow adds workflows.
Full-text search
tsvector normalizes text and builds an index for fast lookups. pg_trgm uses trigrams for partial and fuzzy matches. ParadeDB layers BM25 ranking on top for richer search.
Document store
JSONB stores JSON as binary and lets you index it, so you keep schemaless data. FerretDB offers a MongoDB-compatible API on Postgres.
Vector search / AI
pgvector adds a vector type and nearest-neighbor search for embeddings. pgvectorscale strengthens the index and holds speed at scale.
Time-series data
TimescaleDB splits tables automatically to speed writes and aggregates. pg_partman manages partitions by time or range.
Analytics / OLAP
pg_analytics speeds analytical queries with columnar storage. DuckDB integration lets you summarize and read files straight from Postgres.
Graph database
Apache AGE queries graphs with openCypher. Recursive CTEs (WITH RECURSIVE) walk hierarchies and graphs without an extension.
Geospatial
PostGIS adds spatial types, functions, and indices. You compute distance and intersection in plain SQL.
Conclusion
Every extra store adds deployment, monitoring, and incident response. Check whether Postgres is enough first, and your stack stays far simpler.
