This document discusses monitoring Docker containers and applications. It outlines the challenges of monitoring at the container, application, and infrastructure levels. An approach is proposed that uses agents running on each host to collect data from drivers for the host, containers, and applications. The data is sent asynchronously to a central monitoring controller which aggregates, stores, analyzes and derives insights from the data. Predictions, usage patterns, and capacity planning information can be determined to help optimize container resources and performance.