Getting Started

Predictive Pod Autoscaler (PPA) is an AI-driven horizontal pod autoscaling solution for Kubernetes that proactively scales workloads based on historical usage.

Prerequisites

Before deploying PPA, ensure your cluster meets the following requirements:

• Kubernetes: v1.25 or newer
• Helm: v3.x
• KEDA: Enabled for horizontal pod autoscaling
• Argo Workflows: Required for model training orchestration
• KServe: Required for serving predictive models
• Kro: Required to manage the Model Custom Resource
• Prometheus: Source for historical metrics
• S3-Compatible Storage: (e.g., SeaweedFS, Minio) for storing model artifacts

Next Steps

Follow the guides below to get PPA up and running:

• Installation Guide
• Usage Guide