What is Kubernetes sessions - 6: Observability

This blog delves into the core pillars of Kubernetes observability - logs, metrics, and traces - and how they collectively contribute to system insight. We explore essential monitoring tools such as Prometheus, Grafana, and Jaeger, shedding light on their functionalities.

Observability is a crucial aspect of managing and monitoring modern distributed systems. In the context of Kubernetes, it plays a significant role in ensuring the performance, reliability, and overall health of containerized applications. 

Observability Pillars

In the context of Kubernetes, observability relies on three main pillars: logs, metrics, and traces.


Logs are textual records of events that occur within the system. They provide valuable insights into application behavior and help identify issues. In Kubernetes, logs can be generated by the platform itself (platform logs0, as well as by individual containers and applications (app logs). 


Metrics are numerical values that represent the state of a system over time. They can help track the performance, resource usage, and health of Kubernetes components, as well as the applications running on them. Examples of metrics include CPU usage, memory consumption, and network latency.


Traces provide a detailed view of individual requests as they flow through a distributed system. They help identify bottlenecks, diagnose issues, and optimize performance. In Kubernetes, traces can be collected from various components, such as platform apps as well as business apps.

Logs, metrics and traces generated by business apps are usually the responsibility of the teams that built those apps (and have the knowledge), while platform logs should be the responsibility of the platform engineers.

Monitoring Tools and Techniques

There are several popular monitoring tools available for Kubernetes, including Prometheus, Grafana, and Jaeger.


Prometheus is a widely-used monitoring system and time-series database. It is designed for reliability and scalability, making it well-suited for Kubernetes environments. Prometheus collects metrics from instrumented targets by scraping HTTP endpoints.


Grafana is a visualization and analytics platform that integrates with various data sources, including Prometheus. It allows users to create customizable dashboards to visualize and analyze the collected metrics.


Jaeger is a distributed tracing system that can help track request flows across services in Kubernetes environments. It provides valuable insights into latency, errors, and dependencies between services.

In addition to these tools, there are various techniques for implementing observability in Kubernetes, such as using sidecar containers for log and metric collection, and leveraging DaemonSets for deploying monitoring agents on each node. This choice can be dependent on requirements such as avoiding DaemonSets with wide system access.

Best Practices for Observability in Kubernetes

Define Clear Objectives

It's essential to establish clear objectives for observability to ensure that the collected data is actionable and useful. Objectives should be based on the desired outcomes for application performance and reliability.

Golden Signals

The concept of "golden signals" in Kubernetes refers to a set of key metrics that help provide a high-level view of system health. These may include request rate, error rate, latency, and saturation. Monitoring these signals can help quickly identify issues and guide further investigation.

Alerts and Dashboards

Setting up alerts based on thresholds for critical metrics can help detect issues early and notify the responsible teams. A good platform should have adapters for the most common alerting endpoints, such as OpsGenie, PagerDuty, Slack. Alerts should contain links to dashboards to be able to quickly drill down to potential causes.

Monitoring dashboards should be designed to provide a clear view of the system's health and performance, focusing on the most relevant metrics for the target audience. A good platform makes it easy to find the right dashboard, offering shortcuts to audience specific interests.


Leveraging automation can significantly improve the efficiency of observability efforts in Kubernetes. This may involve automating the deployment of monitoring tools, configuration updates, and the collection and processing of data. As more tools are using A.I. we can expect to see more focus on actionable data and suggestions that operators can approve or build out.


In summary, observability plays a critical role in ensuring the performance and reliability of Kubernetes environments. By understanding the core pillars of observability, leveraging popular monitoring tools, and following best practices, teams can gain valuable insights into their Kubernetes deployments. Implementing observability in your Kubernetes environment can help improve application performance, increase reliability, and enable quicker issue resolution. We encourage you to explore these concepts further and apply them to your own Kubernetes setups for enhanced system monitoring and management. We also suggest keeping an eye out in this space for A.I. solutions as anomaly detection is very well suited for it. 

Latest Articles

Navigating the Evolution: Trends and Transformations in Kubernetes Platforms for 2024

Navigating the Evolution: Trends and Transformations in Kubernetes Platforms for 2024

As we look ahead to 2024, the excitement around building and managing container and Kubernetes platforms is shifting to a more realistic outlook. Companies are realizing that these tasks are more complex than originally thought. In the bigger picture, we can expect things to come together and simplify in the coming year. Let's break it down.

Read more
Mastering Dockerfile USER

Mastering Dockerfile USER

Mastering Dockerfile USER: The Key to seamless Kubernetes Deployment

Read more