Question

A media company is building a new video-on-demand platform. Users upload raw video files that must be transcoded into multiple resolutions before publishing. The transcoding jobs are CPU-intensive, can run for 20-40 minutes each, and volume is highly variable (near zero overnight, thousands of jobs during peak hours). The engineering team wants a fully managed solution that scales automatically, minimizes operational overhead, and does not require them to manage a fleet of servers or handle interruption logic. Which approach best meets these requirements?

Accepted Answer

Use AWS Elemental MediaConvert with jobs submitted via a queue, letting the service scale transcoding capacity automatically based on demand.

Answer

Deploy an Auto Scaling group of Spot Instances running FFmpeg, with an SQS queue and target-tracking scaling on queue depth, plus custom logic to re-queue jobs interrupted by Spot reclamation.

Answer

Run transcoding on a fixed fleet of large On-Demand EC2 instances sized for peak load, using cron to pull jobs from an S3 bucket.

Answer

Use AWS Lambda functions triggered by S3 uploads to perform transcoding, scaling concurrency automatically with upload volume.