Cloud Video Production on AWS: Instances, Costs & Best Practices

AWS gives you the building blocks to run full live production in the cloud—contribution, switching I/O, transcoding, packaging, and delivery. Here’s how to design a stack you can trust and afford.

Pick the right instances

• Control & routing: c7g/m7g for APIs, signaling, and light I/O.
• Transcoding: g5-class (GPU) for H.264/HEVC density; vt1-class (U30 FPGA) for efficient live ladders.
• Storage & packaging: S3 for origin, CloudFront for scale, with EBS/instance store for scratch.
• Networking: Prefer enhanced networking (ENA), same-AZ placement, and dedicated security groups for SRT/NDI.

Cost pattern to expect

• Compute: Pay for real-time transcode capacity; consider Spot or Savings Plans for predictable events.
• Data transfer: Egress costs matter. Use CloudFront and regional edges to optimize.
• Storage: S3 Standard for hot, Lifecycle to IA/Glacier for archives. Keep origins small; rely on CDN cache.

Reference workflow

1. Ingress: Field encoders send SRT to public listeners in two AZs.
2. Processing: Transcode to ABR ladders (H.264/HEVC), generate captions/thumbnails if needed.
3. Packaging: HLS/DASH origin on S3; signed URLs if content is premium.
4. Delivery: CloudFront with appropriate TTLs and geo settings. Add a backup distribution.
5. Monitoring: Metrics, logs, and health checks feeding alerts and dashboards.

Security & reliability

• Lock down SRT ports, use passphrases, and rotate credentials.
• Multi-AZ by default. Plan single-click failover for pipelines and origins.
• Restrict IAM roles to least privilege; audit logs and config regularly.

NDI and interop

When you need NDI in the cloud, keep adapters, discovery, and mixers in the same AZ/VPC. Use private subnets and peering for hybrid onsite-cloud networks.

Runbook for show day

• Warm up instances and encoders; confirm bitrates and ladders.
• Test main/backup SRT paths and DNS failover.
• Record ISOs; verify storage headroom and write speeds.
• Have a rollback plan and a known-good profile you can switch to fast.

Outcome: A lean, resilient AWS setup that scales with your audience and protects your budget.