Configuration Command Center
Enter, validate and version the environment information used directly by generated deployment scripts.
Automatic tier transitions
These rules control how sites automatically climb (LIU → DIA → RIU → SIU) or step down based on observed accelerator utilization and request mix. They apply to tenant texas-testbed (from the Environment tab).
When off, the engine still measures utilization and shows recommendations but never switches a site automatically.
Upgrade rules
Sites upgrade one tier when sustained GPU utilization exceeds the threshold, or the accelerator (GPU/TPU) share of the request mix is dominant; heavy TPU or distributed-training load escalates straight to SIU.
Downgrade rules
Sites step down one tier when accelerator utilization stays below the low threshold while the request mix is CPU-dominant.
SIU load types
Comma-separated load types that force escalation to the strategic SIU tier.
Pacing
The rolling window used to average telemetry, plus cool-downs that stop a site ricocheting across tiers and keep user overrides in place.
Per-tenant capacity quotas
Quotas cap the aggregate accelerator footprint tenant texas-testbed (from the Environment tab) can hold across all capacity units. They are enforced when reservations are placed; tenants without a quota are unlimited.
Current usage
All configured quotas
| Tenant | Max accelerators | Per request | Max reservations | In use | Updated by |
|---|---|---|---|---|---|
| Open this tab to load quotas. | |||||
Governed provisioning workflows
Governed simulation. These workflows run real multi-step state machines with hash-chained evidence for TPU provisioning, air-gapped registry bundles and multi-cloud / edge-to-cloud orchestration — but no actual cloud, TPU or registry infrastructure can be provisioned from this environment. Use the generated evidence to rehearse and audit the process before running it against real infrastructure.
Workflows apply to tenant texas-testbed (from the Environment tab).