Performance envelope¶
Status: Available in ETLantic 0.41.0. Coordination microbenchmarks only— not warehouse throughput or SLA claims.
Residual evaluation lead¶
| Topic | 0.41 |
|---|---|
| Maturity | Beta |
| Support | Community; no SLA |
| Scale claim | Measured framework overhead envelopes only |
| Engine sizing | Adopter-owned (measure Polars/Pandas/SQL/Spark yourself) |
| Multi-tenant / HA | Not included in 0.37; multi-tenancy is planned first-class, HA claims remain separately gated |
What ETLantic claims¶
ETLantic publishes measured p95 envelopes for modeling, validation, planning,
plugin discovery, and interchange reconciliation. CI fails when those budgets
regress. See Performance guidance and
benchmarks/baselines/core.json.
What ETLantic does not claim¶
- Rows/second or warehouse sizing for Polars, Pandas, SQL, or Spark
- End-to-end SLA, capacity, or noisy-neighbor guarantees
- That coordination p95 predicts production pipeline wall time
How to measure for your pilot¶
- Run
uv run python scripts/check_benchmarks.pyfrom a checkout (optional). - Measure your own graph on the chosen engine with representative data.
- Include failure, cancellation, and interchange paths from the resilience work (0.23+)—not only happy-path throughput.
- Record accepted residual risks in your deployment review (Production readiness).