Run millions of batch jobs across many Kubernetes clusters — fairly and at high throughput — with Armada.

Welcome to Armada: Multi-Cluster Batch Scheduling at Scale.

A practical, in-depth guide to Armada, the CNCF multi-cluster batch job scheduler from G-Research. Learn how Armada pools many Kubernetes clusters into one fair, high-throughput batch system: the global scheduler and pool-wide view, per-cluster executors, queues and weighted fair-sharing, priorities and preemption, multi-cluster placement, the event-driven Pulsar-backed architecture for massive scale, and operating Armada in production — for quantitative, scientific, ML, and large-scale data batch workloads.

This title is part of the ShriIra library and is free to read in full, right here — our small contribution to making world-class knowledge easy to reach.

A note on reading it: open the Contents menu at the top of the reader to jump between chapters, use the Aa menu to set a comfortable text size, theme (light, sepia, or night), and single- or two-page layout. Your place is saved automatically, so you can always pick up where you left off.

We hope it serves you well.

— Shriira Press

Contents

1. Chapter 1 — What Armada Is
2. Chapter 2 — Batch Scheduling and Multi-Cluster
3. Chapter 3 — Armada Architecture
4. Chapter 4 — Jobs and Submission
5. Chapter 5 — Queues and Fair Sharing
6. Chapter 6 — Multi-Cluster Scheduling
7. Chapter 7 — The Executors
8. Chapter 8 — High Throughput and Scale
9. Chapter 9 — Operations and Observability
10. Chapter 10 — Armada in Practice