# Administration — Platform Operations

![Administration — Platform Operations](/docs-content/ce5dc7b8cd-06-platform-operations.png.webp)

**Navigation:** Administration → Platform Operations  
**Access:** platform-operations permission  
**Primary audience:** platform operators, site reliability teams, and authorized incident responders

## Purpose

Platform Operations monitors application and processing reliability across ingestion, media, transcription, analysis orchestration, indexing, exports, queues, storage, and compute. It supports incident detection, capacity decisions, quarantine, retry, and recovery.

## Scope controls and indicators

Period selects the operational window. Environment chooses Development, Testing, Staging, or Production where permitted. Service filters the subsystem. Refresh controls automatic or manual metric update.

Availability shows the selected service objective period. Queue Depth counts pending work and high-priority items. CPU or relevant resource utilization shows current capacity. Quarantined Jobs identifies work requiring operator review.

## Service health and throughput

The service list shows operational, degraded, unavailable, or maintenance state and a relevant throughput or backlog measure. The timeline shows movement during the selected period. **Open system health** provides service-level detail.

## Queue and operator attention

**Queue composition** separates transcription, analysis, indexing, exports, or other work. **Operator attention** lists incident or quarantined-job ID, concise cause, severity, age, and state. Opening an item displays affected records, retries, errors, and safe next actions.

## Supporting panels

**Capacity headroom** reports remaining CPU, GPU, storage, and worker capacity using defined thresholds. **Recent incidents** shows current and recent events. **Operator controls** provides retry, scale, and incident actions.

## Actions

- Open system health, queue, job, or incident details.
- Retry eligible jobs after resolving or acknowledging the cause.
- Scale workers within approved limits.
- Quarantine or release an item.
- Open, update, or close an incident.
- Change refresh or service scope.

## Recommended procedure

Prioritize unavailable services and high-impact backlogs. Diagnose the cause before retrying. Use correlation IDs to inspect affected jobs, then choose retry, quarantine, scaling, or incident response. Confirm recovery and backlog drain before closure.

## Boundaries

Model quality gates and AI release decisions belong to [AI Operations](/documentation/packages/bnd-aiops/screens/administration-ai-operations). Connector credentials and source mappings belong to [Sources & Integrations](/documentation/packages/bnd-evidence/screens/administration-sources-and-integrations). Security events belong to [Audit & Security](/documentation/packages/bnd-aiops/screens/administration-audit-and-security).

## Related screens

[AI Operations](/documentation/packages/bnd-aiops/screens/administration-ai-operations), [Sources & Integrations](/documentation/packages/bnd-evidence/screens/administration-sources-and-integrations), and [Audit & Security](/documentation/packages/bnd-aiops/screens/administration-audit-and-security).