Fundamentals of DevOps and Software Delivery

Chapter 10. How to Monitor Your Systems

In Chapter 9, you learned how to store, query, replicate, and back up your data. That chapter focused primarily on data about your customers, such as user profiles, purchases, and photos. This chapter focuses primarily on data that gives you visibility into your business, or what is typically referred to as monitoring.

In the Preface, you heard about LinkedIn’s software delivery struggles. We struggled with monitoring, too. We collected metrics and logs, but the tools to understand that data were unusable, so we were often flying blind, and bugs and outages could go unnoticed. In 2010, an intern created inGraphs, a UI for visualizing our metrics. It had a profound impact on the company, as suddenly we could spot problems earlier and understand what users were doing. We overhauled monitoring even more as part of Project Inversion, and before long, inGraphs was on screens all over the office. As David Henke, LinkedIn’s senior vice president of engineering and operations, liked to say: if you can’t measure it, you can’t fix it.

So, what should you measure? The following four items are the most commonly used monitoring tools and techniques:

• Logs

• Metrics

• Events

• Alerts

This chapter dives into each of these topics, and as you go through them, you’ll try out examples, including using structured logging in Node.js, creating a dashboard of EC2 metrics in Amazon CloudWatch, and configuring Route 53 health checks with alerts to notify you ...