QuestDB alternatives: test time-series ingest and late data

For QuestDB alternatives, use real-shaped time-series records. The deciding factor may be how the system handles delayed or out-of-order observations, not headline rows per second.

Benchmark

1. Ingest timestamps, tags, measurements, duplicates, and late arrivals at expected volume.
2. Query recent windows, long-range aggregates, and high-cardinality filters.
3. Measure write throughput and query p95 simultaneously.
4. Test retention, downsampling, partitioning, schema changes, and recovery.
5. Verify timestamp precision, timezone handling, export, and operational cost.

Keep raw sample data and expected results to compare engines fairly. A fast insert benchmark that ignores query freshness is incomplete.