Observability & Monitoring in SRE Practices

Observability & Monitoring in SRE Practices

Observability and monitoring – two different but inseparable parts of SRE. Observability allows SREs to understand the system, and monitoring is to watch the system in different states. Teams need to use and practice both terms to understand the actual state of the system. This article talks about the differences between Observability and monitoring and how to measure them.

Difference Between Observability And Monitoring


Observability is a technical process of understanding the outputs of a system. Using it, SRE, developers, and operators can debug their systems. It can also identify new patterns and properties at an early stage. In some cases, the code behavior varies in staging and production environments. Proactively observing the system can reduce the negative impact. The three critical areas of Observability to learn what’s going on with the code in production are:

1.      Logs: Output of the code is called logs. It includes immutable, time-stamped records in a system. It has no specific standard.

2.      Metrics: The aggregated data of the performance is called metrics. It refers to the single data that is monitored over time. Both DevOps and SRE teams often watch and fix metrics.

3.      Traces: Visualizing the operation state, from the parent event to the child event, is called traces. It includes information like start time, ending time, duration, parent-id, child-id, etc.


Monitoring focus on finding issues and alerts in the performance to improve end-user experience. In other words, monitoring is finding the symptoms and causes. There are two types of monitoring.

1.      White Box: It shows the insight of the various parts by the internals like logs, HTTP handlers, interfaces, etc. Using white-box, SRE teams can find the causes and symptoms of the issues.

2.      Black Box: It shows the behavior of users. It includes error responses and latency from the users’ perspective. It is more focused on the symptoms.

Measuring Observability and Monitoring

Using Observability and monitoring tools, SREs can find alerts and improve processes over time. Members can fix some issues manually. But critical alerts require tools to fix. SREs need to instrument the code to emit logs and metrics. Members can also automate the alert process.

By measuring Observability and monitoring, members can find changes to monitoring configuration, handling and distributing alerts, numbers of actionable and silenced alerts, MTTD (Mean Time to Detect), MTTR (Mean Time to Resolution), and usability, etc. Based on the findings, organizations understand whether the systems are working efficiently or not.

Businesses can use monitoring tools like DataDog, Nagios, Grafana, Prometheus, etc., and Observability tools like Splunk, Open Telemetry, Google Cloud’s Operations Suite, Honeycomb, etc. No particular team is responsible for observability and monitoring. SRE and DevOps teams are both responsible for implementing them.

The goal of observability and monitoring is to improve organizations’ ability to deliver quality software quickly and sustainably. Businesses that are serious about achieving this goal need to focus on improving their monitoring systems.