From Chaos To Clarity: Scaling Observability at Dropbox With Centralized Logging Solution

This conference talk explores how Dropbox tackled the massive challenge of scaling observability for systems generating terabytes of logs daily. Learn about their journey from a chaotic process where developers had to access logs by logging into individual servers to implementing a centralized logging solution. Discover how container adoption complicated matters further as logs would disappear when containers terminated, necessitating a scalable, persistent solution using open-source tools. The speaker details Dropbox's implementation of a robust observability framework built around Loki as their logging solution, including the extensive optimizations required to handle Dropbox's enormous data volume while maintaining reliable and efficient query performance. The presentation covers deployment strategies, challenges encountered, and techniques for achieving high-performance logging at scale. Additionally, learn how Dropbox integrated Grafana to create a unified view of logs and metrics, significantly enhancing troubleshooting capabilities and security monitoring. Gain valuable insights from Dropbox's experience scaling observability with open-source solutions that can be applied to your own infrastructure challenges.