Chaos engineering and scalability at Audible.com

Description

You think you know how your system works, until 1:00am one night when it just doesn’t. No matter how big or small your system is, the complexity between dependencies will become chaotic when you least expect it. Chaos engineering is the way to understand and tame this chaos.

As Audible.com continues on a path toward distributed microservices and serverless technologies, the audio and playback experience team has utilized chaos engineering to experiment and measure resiliency and reliability of the playback experience. Tyler Lund discusses why Audible is evolving its architecture toward microservices and serverless architectures and how the customer experience has improved through measured experimentation with chaos engineering. With chaos engineering, the goal is to experiment with failure scenarios within software systems to verify and test the resiliency of a large system. In a distributed system at Amazon scale like Audible, a single point of failure can affect millions of users.

Join Tyler to learn more about how Audible tests disruptions and failures like data center outages, network latency, host failures, massive requests, and others to ensure the customer experience never suffers. Tyler dives deep into Audible’s reasons for embracing chaos engineering, how it has been applied and automated, the types of experiments run, and how Audible garnered support throughout the organization to expand experimentation. He also covers the metrics and goals used to get executive support and track success along with specific tests the team runs, how they built an automation framework, and the issues they’ve uncovered.

Learn how Audible embraces chaos to make things better for its users and how you can too.