Monitor everything

This document discusses the importance of comprehensive monitoring for containerized applications and infrastructure. It recommends monitoring at multiple levels, from the host and operating system to the network, orchestration layer, and applications themselves. The document outlines common failure models for each of these levels. It also provides an example of how detailed metrics and monitoring could have helped identify issues that caused an e-commerce site outage. The document concludes by recommending best practices for monitoring, such as starting small, avoiding alert fatigue, and testing monitoring systems.