Book Image

Observability with Grafana

By : Rob Chapman, Peter Holmes
Book Image

Observability with Grafana

By: Rob Chapman, Peter Holmes

Overview of this book

To overcome application monitoring and observability challenges, Grafana Labs offers a modern, highly scalable, cost-effective Loki, Grafana, Tempo, and Mimir (LGTM) stack along with Prometheus for the collection, visualization, and storage of telemetry data. Beginning with an overview of observability concepts, this book teaches you how to instrument code and monitor systems in practice using standard protocols and Grafana libraries. As you progress, you’ll create a free Grafana cloud instance and deploy a demo application to a Kubernetes cluster to delve into the implementation of the LGTM stack. You’ll learn how to connect Grafana Cloud to AWS, GCP, and Azure to collect infrastructure data, build interactive dashboards, make use of service level indicators and objectives to produce great alerts, and leverage the AI & ML capabilities to keep your systems healthy. You’ll also explore real user monitoring with Faro and performance monitoring with Pyroscope and k6. Advanced concepts like architecting a Grafana installation, using automation and infrastructure as code tools for DevOps processes, troubleshooting strategies, and best practices to avoid common pitfalls will also be covered. After reading this book, you’ll be able to use the Grafana stack to deliver amazing operational results for the systems your organization uses.
Table of Contents (22 chapters)
1
Part 1: Get Started with Grafana and Observability
5
Part 2: Implement Telemetry in Grafana
10
Part 3: Grafana in Practice
15
Part 4: Advanced Applications and Best Practices of Grafana

Exploring metric types and best practices

Metrics, along with logs, are an essential tool for software developers (Diego) and operators (Ophelia), providing them with indicators regarding the state of applications and systems. Resource usage data is great for monitoring a metric that captures numerical data over time. There are many different types of resources but some good examples would be CPU or RAM usage, the number of messages in a queue, and the number of received HTTP requests. Metrics are frequently generated and easily enriched with labels, attributes, or dimensions, making them efficient to search and ideal in determining if something is wrong, or different from usual.

A metric commonly has the following fields:

  • Name: This uniquely identifies the metric
  • Data point value(s): The data that’s stored varies by metric type
  • Dimensions: Additional enrichment labels or attributes that support analysis

Metrics capture the behavior of the data they...