Zone 1 · Model & scenario
Simulation controls
Every run is a deterministic what-if projection from seed + assumptions. Nothing here measures production.
Base URL of the ShadowBox API — the live Worker or a local serve.
8 chaos cards. Same seed + scenario always replays the same result.
Integer, deterministic.
API unreachable — simulation not started.
Could not reach —. Check the API base, verify network access, then retry. Previous results are preserved below.
Zone 2 · Verdict vs baseline
No run yet — press Run
Simulated deltas will appear here with seed and assumptions attached.
What-if estimate only. Do not treat as production measurement. See assumptions in Zone 7.
Zone 3 · KPIs
Result indicators
● EMPTY — no dataError rate
—
awaiting run
p99 latency
—
awaiting run
Throughput
—
awaiting run
Cascade depth
—
awaiting run
Metrics hash
—short hash of simulated metrics
Zone 4 · Topology
Architecture graph
Healthy (icon + label)
Degraded (icon + label)
Failing (icon + label)
Simulated edge
100%
Select a node for detail (keyboard: Tab + Enter).
Node detail — select api, cache, or database
Simulated per-node internals for the current run. Estimates only.
Utilization
—
Queue depth
—
Timeouts
—
Fails
—
Zone 5 · Metrics
Latency distribution
Per-component table (simulated)
| Component | Util | Queue | Timeouts | Status |
|---|---|---|---|---|
| No run yet — empty state. Press Run to simulate. | ||||
Zone 6 · Compare
Multi-scenario matrix
All rows are simulated with the current seed. Breaches vs baseline are highlighted with icon + text, never color alone.
| Verdict vs baseline | |||||
|---|---|---|---|---|---|
| Empty — run all scenarios to populate the matrix. | |||||
Zone 7 · Guide
Tour, catalog & assumptions
Guided tour — 5 steps
Step 1 of 5
Chaos-card catalog — expected outcomes (simulated)
Metrics glossary
- error_rate
- Simulated share of failed requests, 0–1. Breach if > 0.01 above baseline.
- p99
- Simulated 99th-percentile latency in ms. Breach if > 2× baseline.
- throughput
- Simulated successful req/s the model sustains.
- cascade_depth
- Simulated max failing hops from entry (0–3).
- metrics hash
- Short deterministic fingerprint of the simulated metric set + seed.
- verdict
- PASS or REGRESSION vs baseline thresholds. Estimate, not a test result.
⚠ Assumptions — read before quoting results.
- Model: api → cache → database, fixed topology, configured capacity/latency/timeout per component.
- Seeded server-side RNG; same seed replays an identical metrics hash.
- Confidence is always low; outputs are what-if estimates, never production measurements.
- Thresholds: error +0.01, p99 ×2, throughput −20%, cascade ≥ 2 ⇒ REGRESSION.