Groupon runs Kubernetes clusters on thousands of hosts for its engineers, auditors and automated systems. Traffic from millions of Groupon customers flows through Kubernetes. Taking downtime for any reason, especially self-inflicted, is detrimental to the business' brand and bottom line. To provide zero downtime during platform upgrades, a new cluster is created, validated and traffic is migrated slowly to it. In addition to Kubernetes releases, platform upgrades include any change made to key software components or the underlying cloud infrastructure it runs on. Join Asaf and Jonathan of Groupon as they discuss architectures for continuously executing zero-downtime platform upgrades.
DevOps World 2021