TechPrufer
Get started
Netflix
Netflix logoNetflix · Feature trail

Batch Compute Management

2 posts · Dec 2024 – Jun 2026

What this trail covers

Netflix has evolved its batch compute management by migrating from a custom-built solution (CMB) to an open-source Kubernetes-native system, Kueue. This transition aims to simplify batch job execution, enhance resource utilization through fair sharing and preemption, and leverage the Kubernetes ecosystem for greater flexibility and innovation. The migration involved integrating Kueue into their existing container platform, Titus, and re-architecting tenant and capacity management to align with Kueue's concepts of ClusterQueues and LocalQueues. The system now manages millions of batch workloads, with improved fair sharing and preemption leading to increased resource utilization.

All Netflix trails
  1. How Netflix Simplified Batch Compute with KueueThis post details the migration of Netflix's custom batch compute management system (CMB) to Kueue, an open-source Kubernetes-native job queueing system. The migration involved replacing CMB's queuing and scheduling logic with Kueue, integrating it into the Titus container platform, and re-architecting tenant management to map to Kueue's Cohorts, ClusterQueues, and LocalQueues. Key technical challenges addressed included maintaining API parity, migrating complex use cases early, and tuning Kueue for high QPS and throughput. The post highlights the benefits of Kueue, such as preemption and fair sharing, leading to improved resource utilization and faster turnaround times for critical workloads.
  2. MediumThis post details the migration of Netflix's batch compute management from a custom-built solution (CMB) to Kueue, an open-source Kubernetes-native system. It highlights the technical challenges and solutions involved in integrating Kueue into their existing container platform (Titus), including adapting tenant and capacity management to align with Kubernetes principles. The post emphasizes the benefits of this migration, such as simplified batch job execution, enhanced resource utilization via fair sharing and preemption, and leveraging the Kubernetes ecosystem.