Observability 101

Observability isn't just about gathering metrics.

In this article, I will explain the essentials of observability, why it matters, and how to get started.

I started my career almost a decade ago with an internship in infrastructure, but as soon as I finished, I started to work in an event management and monitoring position, where I was exposed to IBM Tivoli Monitoring and Netcool Omnibus.

This role was my first real jump into the world of observability, and I quickly saw the importance of being able to anticipate and respond to system issues before they impacted users and what were the consequences when observability wasn’t up to par — believe me they are severe.

But enough about me, let’s jump into the goods:

What Is Observability?

Observability is a mechanism that can measure a system’s internal state by examining its outputs.

Traditionally, monitoring involved setting up alerts for specific metrics, such as CPU usage or memory, but observability goes deeper.

Observability aims to provide a more comprehensive view of your system’s health, performance, and reliability by focusing on three main pillars:

1. Metrics — Data measurements over time such as request rates, error rates, or system resource usage (CPU, Memory, etc)

2. Logs — Text records of everything that is happening inside your application or system that can be parsed to get detailed insights into events

3. Traces — Give you the ability to track the entire journey of your requests across services, showing how individual components interact.

These pillars work together to provide a holistic system view, enabling DevOps teams to detect and resolve issues more effectively.