Skip to main content

Deployment Topology

:::info Source Sourced from services/media-service/DEPLOYMENT_TOPOLOGY.md in the documentation repo. :::

1. Containers​

  • media-api — REST API.
  • media-scanner — AV + content-safety workers.
  • media-transcoder — ffmpeg / MediaConvert orchestrator.
  • media-ai-worker — AI image/TTS/STT generation.
  • media-outbox-relay.

2. Scaling​

ContainerMinMaxHPA
api315CPU>60%
scanner330queue > 50
transcoder350queue > 20
ai-worker215queue > 30

3. Resources​

api: 500m/2000m, 512Mi/1.5Gi. transcoder: 2000m/8000m, 4Gi/16Gi (CPU-heavy).

4. Storage​

  • S3 per region; per-tenant prefix.
  • Hot + Glacier archival.
  • MediaConvert (AWS) or equivalent managed transcoder offload optional.

5. CDN​

  • CloudFront / Fastly for HLS/DASH + thumbnails + captions.
  • Signed URLs enforce per-asset + per-user scope.

6. Regional​

  • Per region. Assets pinned to tenant homeRegion.
  • CDN anycast globally.

7. Service Mesh​

mTLS. Egress: ai-gateway, authoring, content, search, analytics.

8. Release​

Blue/green API. Rolling workers.

9. DR​

RPO 15 min (S3 versioning). Assets irrecoverable if S3 object lost (unless originals backed up).