Maintaining a high-quality playback experience for users is extremely critical whether it is for on-demand content or for large-scale live events. Origin failures occur when a server is overwhelmed with requests, due to congested networks or other unexpected events. These failures can keep origin responses from getting back to the client; hence, the need for redundant server architectures.
As part of Akamai's mission to provide easy-to-implement broadcast-quality streaming over the internet, we've released Origin Failover behaviors for our Adaptive Media Delivery (AMD) product to make it easier to detect and recover quickly from such failures. Origin Failover offers a self-service solution for maintaining continuity of playback in the event of an origin failure. Origin Failover behaviors are available now in Property Manager for your AMD configuration.
How it works
With Origin Failover, you can take advantage of Akamai's role in the delivery of your streaming content by detecting problems upstream from your users. Origin Failover is designed to identify origin and network issues before they extend to the client, helping to reduce the likelihood of a rebuffer event during an origin failure. By detecting origin issues as close to the origin as possible, we reduce the time to recovery by failing over before the client-side player detects an issue. This minimizes impact to playback performance in terms of rebuffering during failures.
Additionally, when detecting origin issues in network, we reduce the number of end user requests to a failing origin, reducing load when it is needed most. Requests can be routed to a backup origin without the delay of issuing a client side redirect, while also issuing a custom response code to the client, thereby offering continued playback with the ability to report on the failover event not only within Akamai's reporting but also your clientside player metrics.
Origin Failover supports failure events during startup and midstream for both Live and VOD playback. (Note: live requires time synced primary and backup origins)
Turning it on
Origin Failover consists of two behaviors in Property Manager that implement failure detection and failover actions. Failure detection is configured by the behavior "Origin Failure Recovery Policy" where failure scenarios can be triggered by HTTP response codes, forward connect timeouts, or detected slow connections.
Failover actions are configured by the "Origin Failure Recovery Methods" behavior and offer the option of trying an alternate origin or responding with a custom status code back to the client.
Turning the feature on is a two step process.
Step 1: Add the "Origin Failure Recovery Policy" Behavior to your default rule or streaming sub rule.
When you enable the feature you will be presented with additional options for detecting and responding to the origin failure. Some options will be highlighted in red and will require user input. To learn more about these options, please visit our guide on configuring your recovery policy.
Here you will define the "Recovery Method Configuration Name" that will be used as the trigger for the action you'll take in the event of a failover trigger.
Step 2: Create a Rule with a Criteria of "Recovery Configuration Name" and a Behavior of "Origin Failure Recovery Method". Note: If retrying with an alternate origin, you'll want to add the "Origin Server" Behavior under this rule as well so you can configure the details for the alternate origin.
If you'd like to give Adaptive Media Delivery a trial run, please sign up for a free trial.