HPC Workload Scheduling, with Ricardo Rocha

In this episode of the Kubernetes Podcast from Google, hosts Abdel Sigiwar and Mofi Rahman interview Ricardo Rocha, the platform infrastructure team lead at CERN, about using Kubernetes for scientific computing. The episode begins with news items, including the introduction of Node Feature Discovery (NFD) in Kubernetes, Google's Gemini CLI, the CNCF's Vietnamese cloud native glossary, and the appointment of Jonathan Bryce as the new CNCF executive director. The interview covers Ricardo's interest in aviation and draws parallels between flying and managing Kubernetes clusters. Ricardo discusses CERN's transition to cloud native technologies to handle large computing resource requirements, the differences between scientific and traditional IT service workloads, and the evolution and benefits of using Kube for batch scheduling, fair sharing, and resource optimization in on-premises data centers. The conversation also explores the future of batch workloads, multi-cluster management, and the integration of Slurm with Kubernetes, highlighting the importance of community involvement and feedback in the Kubernetes ecosystem.