"KEDA Event-Driven Autoscaling: Scale Kubernetes on Queues, Streams, and Metrics"
Modern Kubernetes autoscaling is easy to demo and hard to operate. This book is written for experienced platform engineers, SREs, and senior application developers who need autoscaling that behaves correctly under real-world queues, streams, bursty traffic, and strict SLOs. Instead of treating scaling as a YAML tweak, it frames autoscaling as a systems problem—control loops, signal quality, reconciliation, and failure domains—so you can predict and shape behavior rather than react to surprises.
You'll build a precise mental model of how KEDA composes native Kubernetes primitives: HPA decision mechanics, the aggregated metrics APIs, and the external.metrics.k8s.io request path from HPA to KEDA's metrics adapter and on to the event source. The book then goes deep on ScaledObject design (ownership, lifecycle, multi-trigger behavior, polling/cooldown/activation tuning), trigger and scaler selection, and translating backlog, lag, rate, and concurrency into stable replica counts. Production chapters cover authentication patterns (TriggerAuthentication and cluster-scoped trade-offs), least-privilege RBAC and admission controls, and an operational troubleshooting workflow for metric discovery failures, scale-from-zero gaps, and flapping.
Expect detailed decision criteria, failure-mode analysis, and upgrade guidance anchored in KEDA v2.x. Readers should already be comfortable with Kubernetes controllers, metrics, and workload operations; the emphasis is on advanced configuration, tuning, and safe pr
