TimescaleDB extends PostgreSQL for time-series workloads. Compare alternatives with the same ingest pattern and analytical query, including the old data you must retain.
Load a representative time-window of timestamped records with realistic labels and late arrivals. Measure ingest throughput, p95 query latency over recent and historical ranges, storage size, compression or retention behavior, and backup/restore time. Test a schema change and a continuous or scheduled aggregate if your application uses one. Verify that queries still return correct results across time-zone and daylight-saving boundaries.
Do not benchmark only fresh writes into an empty database. The operational question is whether the system stays predictable as history grows and retention jobs run.
Akousa