Lessons Learned Running GKE Clusters on Spot Instances

Explore the intricacies of running Google Kubernetes Engine (GKE) clusters on Spot Instances in this 21-minute conference talk from SREcon23 Asia/Pacific. Gain valuable insights from Olga Mirensky of Australia and New Zealand Banking Group (ANZx) as she shares lessons learned from managing a fleet of GKE clusters using the cost-effective Spot provisioning model. Discover the hidden complexities beyond the obvious challenge of potential capacity loss, including under-the-hood mechanisms of GKE Spot implementation and edge cases. Learn why strong team collaboration and solid Site Reliability Engineering (SRE) principles are crucial for success in this environment. Understand how leveraging Spot Instances can lead to significant cost savings of up to 91% compared to on-demand prices, and how it integrates with managed Kubernetes services across major cloud vendors.