Niels Claeys, Lead data engineer & partner at Dataminded, explains how his team achieves 97% spot instance usage for their Kubernetes-based data platform. He breaks down the economics behind spot instances for batch processing workloads, where customers can achieve 70-90% cost reduction compared to on-demand instances.

Claeys discusses the strategic approach to balancing cost optimization with reliability requirements. While most workloads can effectively leverage spot instances, he explains how they identify the small portion of critical workloads that require on-demand instances for guaranteed reliability.

Watch the full episode: https://ku.bz/hGRfkzDJW