Q: You deploy a cluster of 50 identical microservices. To ensure zero drifts, they all pull a massive 1GB initial configuration file from a central S3 bucket immediately upon booting via the `CMD` script. Why is this an anti-pattern in container architecture, and what is the immutable alternative?

This brutally violates the principle of Immutable Infrastructure and destroys startup agility. If the S3 bucket goes down, your containers cannot boot. If you deploy 50 pods simultaneously, you abruptly trigger a 50GB spike of completely duplicate network traffic, severely delaying readiness. *Alternative:* Small, rapidly changing configurations should be mounted externally at runtime via Kubernetes ConfigMaps or Docker Swarm Configs (which use fast local tmpfs). If the 1GB file is structurally static (like a machine learning model), it must be baked directly into the Docker image tightly during the CI/CD build phase. The image then acts as an immutable, instant-booting artifact universally across environments.