# Soak Telemetry Final Report (100000)

Conclusion: **looks accelerating**

## Counts
- steps_completed: `100000`
- update_events_total: `7692`
- restarts_total: `7692`
- samples_total: `153`

## Event Latency (ms)
- p50: `212`
- p95: `448`
- p99: `736`
- max: `1362`

## Storage Start/End (bytes)
- event_log_bytes: `23673` -> `3449289`
- sqlite_bytes: `86016` -> `2670592`
- snapshot_bytes: `17456` -> `2482404`

## Slope Metrics
- ms_per_1000_updates: `171.447368`
- ms_per_MB_log_growth: `398.846376`
- ms_per_MB_sqlite_growth: `528.633914`
- ms_per_MB_snapshot_growth: `554.289392`

## Acceleration Heuristics
- avg event_ms first 20%: `133.483871`
- avg event_ms last 20%: `389.290323`
- first-half slope: `29.866667`
- second-half slope: `311.315789`

## Correlation Hints
- corr(event_ms, event_log_bytes): `0.668357`
- corr(event_ms, sqlite_bytes): `0.668035`
- corr(event_ms, snapshot_bytes): `0.668355`

Interpretation: sampled /v1/event latency growth appears accelerating.