What this public report shows
deterministic stream processing profile-based normalization decision classes class-level audit Docker live stream local synthetic benchmark
This public-safe report summarizes a local IDDA Log Governor PoC. It shows how noisy telemetry streams can be reduced into compact, auditable decision classes without storing every raw event as an equal operational fact.
The emphasis is architectural: same decision core, different normalization profiles, different operational granularity.
Telemetry Reduction Results
| Stream / Profile | Input | Retained | Suppressed | Suppression | Decision Classes |
|---|---|---|---|---|---|
| Industrial telemetry — PoC run | 750,000 | 74,750 | 675,250 | 90.03% | 3 |
| Industrial telemetry — Docker live run | 7,500,000 | 750,287 | 6,749,713 | 90.00% | 3 |
| API telemetry — detailed profile | 100,000 | 15,000 | 85,000 | 85.00% | 19 |
| API telemetry — compact profile | 100,000 | 15,000 | 85,000 | 85.00% | 5 |
Public-safe view: raw examples, exact event patterns, class IDs and implementation details are intentionally omitted.
Docker Live Industrial Stream
Docker live industrial telemetry run: Input lines: 7,500,000 Retained lines: 750,287 Suppressed lines: 6,749,713 Suppression ratio: 90.00% Decision classes: 3 Interpretation: A multi-million-event live stream was reduced into a small number of repeated decision classes. The audit remained class-level, not full per-line storage.
Public-safe summary. No raw event patterns or internal class identifiers are shown.
Public-safe architecture summary
Raw telemetry stream
↓
Normalization profile
↓
Repeated event classes
↓
Severity mapping
↓
Decision mapping
↓
Class-level audit
The decision core remains stable.
The profile controls the granularity of the operational view.
Local Throughput Benchmark
benchmark_50M.txt: Lines processed: 50,000,000 Unique classes: 948 Elapsed seconds: 799.9232 Lines / second: 62,506 benchmark_quick.txt: Lines processed: 1,000,000 Unique classes: 948 Elapsed seconds: 15.7713 Lines / second: 63,406
Local synthetic benchmark. Not a production guarantee.
Interpretation
IDDA Log Governor is not positioned as a replacement for observability platforms. It is a deterministic decision layer that can sit before heavier storage, alerting or analytics tools.
Its role is to reduce noisy telemetry into compact, decision-ready signals.
All results shown here are local synthetic PoC results and should not be treated as production guarantees.