Docker packages an application into a self-contained container that includes its dependencies and configuration. This makes deployment behavior consistent across environments, including deployments of AI models and web applications. section 9.6.1
Use Docker when deployment requires more control or consistency than a managed platform provides:
- Encapsulate the application and everything it needs to run in a portable package. section 9.6.1
- Deploy consistently across environments while retaining flexibility for self-hosted infrastructure. section 9.6.1
- Pair with Kubernetes when many containers require load balancing, demand-based scaling, and rolling updates. section 9.6.1
Advanced deployment path
A growing AI web application can move from a streamlined managed deployment to Docker when it needs greater infrastructure control. Kubernetes can then manage the resulting containers as demand and availability requirements increase. section 9.6.1 section 9.7
