Automated Student Data Pipeline with Kubernetes Scaling

Design a robust Kubernetes deployment for a student management microservice that can dynamically scale based on real-time enrollment metrics. Create a Terraform configuration that provisions a multi-zone GKE cluster, implements horizontal pod autoscaling, and includes a Python-based Flask microservice that tracks student registration events. The solution must handle peak enrollment periods with automatic scaling between 10-500 concurrent users, implement comprehensive logging with Prometheus, and include automated rollback mechanisms for deployment failures.