Why Is My Docker Image So Big? A Deep Dive with ‘dive’ to Find the Bloat

Docker images for AI applications often become bloated due to massive library installations and base OS components. The article demonstrates how to diagnose image bloat using Docker's history command and the interactive 'dive' tool to examine each layer in detail. A sample BERT classifier image weighing 2.54GB is analyzed layer by layer, revealing that Python dependencies account for 1.51GB while the base image contributes hundreds of megabytes. The diagnostic approach helps identify specific sources of bloat, enabling targeted optimizations for faster builds, reduced storage costs, and improved security through smaller attack surfaces.