Time-Series (TSDB)
Append-only workloads where time is the primary axis. Optimized for massive write ingestion and chronological roll-ups.
The Append-Only Optimization
If your application generates metrics (CPU usage every 10s), IoT sensor data, or financial tick data, you are dealing with time-series data. Standard RDBMS B-Tree indexes degrade rapidly under massive, continuous insert loads. TSDBs use Log-Structured Merge (LSM) trees or similar structures optimized for appending.
TimescaleDB
Postgres with Superpowers
A brilliant extension for PostgreSQL. It automatically partitions time-series data across time and space (hypertables) while retaining standard SQL compatibility. Ideal if you already know Postgres.
InfluxDB
Purpose-Built
Written in Go (historically) and now Rust for v3. Uses a line protocol for ingestion. Extremely efficient storage engine designed specifically for time-series workloads. Often paired with Grafana.