In a previous Media Blog series, we provided several examples of sophisticated workflows using AWS Media Services. In this post, I demonstrate a workflow that delivers a basic HLS livestream with resiliency for high-visibility events.
This workflow uses multiple levels of redundancy to eliminate several single points of failure that might take a livestream “off the air,” and allows either automatic or near-automatic recovery by the end-user device to restore playback. The self-healing nature of AWS Media Services also means that once a problem in the workflow is resolved, full redundancy is restored.
In this example, a contribution feed is delivered redundantly to two AWS Elemental MediaConnect flows in each of two Regions. AWS Elemental MediaLive encodes the feeds in two Availability Zones, delivering the HLS output to an AWS Elemental MediaStore container in each Region. Finally, the Amazon CloudFront distribution uses an origin failover group to allow the secondary Region to pick up the traffic if there is a loss of stream in the primary Region. Using a combination of epoch locking in MediaLive and properly configured failover settings in CloudFront, we can provide a low or no-impact failover experience for the end user.