KEDA — Prerequisites

What to Know Before Starting KEDA

KEDA is a focused extension to Kubernetes' existing autoscaling model — the prerequisites are mostly about understanding HPA's limitations first, since KEDA's entire value proposition is framed as solving what HPA can't.

Required (Must Have)

1. Kubernetes HPA Fundamentals

bash

kubectl get hpa
kubectl describe deployment

You need to understand what the Horizontal Pod Autoscaler does by default (scaling on CPU/memory) before KEDA's "scale on any metric" pitch means anything concrete — Overview's own framing (a Kafka consumer with low CPU but massive lag) assumes you already know why HPA alone would miss that.

2. Basic Message Queue / Event Source Concepts

•What a message queue is, at a conceptual level (Kafka, RabbitMQ, or similar)

•What "consumer lag" means — messages waiting to be processed

•You don't need deep Kafka/RabbitMQ administration experience, just enough to understand why lag matters as a scaling signal

3. Helm and kubectl Comfort

bash

helm install / helm upgrade
kubectl get pods -n keda

KEDA installs via Helm and is managed entirely through Kubernetes CRDs (ScaledObject, ScaledJob) — the same interaction pattern as any other Kubernetes-native operator.

Nice to Have (Speeds Up Learning)

Prometheus Basics

KEDA can scale on Prometheus query results as a trigger — prior exposure to PromQL (this site's Prometheus section) makes that specific trigger type immediately understandable rather than new syntax.

Cloud-Native Identity Concepts (IRSA, Workload Identity)

Intermediate's TriggerAuthentication material covers both secret-based and cloud-identity-based credential patterns — basic familiarity with your cloud provider's pod-identity mechanism speeds up that section specifically.

HTTP/Networking Basics

The KEDA HTTP Add-on (Intermediate) assumes basic understanding of why a scaled-to-zero HTTP service needs special handling for its first incoming request.

What You Do NOT Need

•Deep Kafka/RabbitMQ administration expertise — KEDA scalers are pre-built connectors, not something you write from scratch

•Prior custom Kubernetes controller/operator development experience

•Multi-cluster infrastructure experience — the core concepts apply on a single cluster

Time Estimate

If you have the prerequisites above:

•Fundamentals: 1 day

•Intermediate + Advanced: 3-4 days

•Job-ready (comfortable with ScaledObjects, activation thresholds, and the HPA behavior tuning covered in Intermediate): 1-2 weeks

Start Here

Go to the Installation section to install KEDA, then proceed to Fundamentals.