DEPLOYMENT GUIDE
Deployment Guide
The governing engine deploys as a single supervisory layer. No application code modification required. No hardware replacement required.
Prerequisites
The governing engine requires access to infrastructure telemetry — CPU utilization, memory pressure, network I/O, and scheduling metrics. No application-level access required. Supported environments: Kubernetes (any version), bare metal Linux, cloud (AWS, GCP, Azure), GPU clusters (NVIDIA H100 and compatible).
Pilot Engagement
All deployments begin with a 90-day pilot engagement. The pilot establishes baseline measurements using the same twin-pair methodology as the V25 benchmark. Baseline is sealed with SHA-256 before the governing engine is activated. This ensures the gain measurement is verifiable and defensible.
Supervisory Layer Activation
The governing engine is activated as a supervisory layer above the existing infrastructure stack. It observes system state, applies the closed-circle governing law, and issues control signals. No application code is modified. No workloads are migrated. The existing stack continues to operate normally.
Measurement & Verification
After activation, the engine runs for a minimum of 30 days before measurement. Results are sealed with SHA-256 and compared against the baseline. The gain is calculated using the same Work-per-Bill equation as the V25 benchmark: (omni_WpB − base_WpB) / base_WpB × 100.
Full Deployment
After pilot verification, full deployment extends the governing engine across all infrastructure layers. The engine scales with the infrastructure — the same governing law applies at every scale, from a single cluster to a multi-region deployment.
EXTERNAL REFERENCES