Prometheus
Definition
An open-source monitoring and alerting toolkit designed for collecting and querying metrics, providing insights into system performance and reliability.
Use Cases
- SoundCloud: Monitor microservices and infrastructure health as the company scaled its service-oriented architecture. — SoundCloud adopted Prometheus early, instrumented services to expose metrics, used PromQL for troubleshooting, and configured Alertmanager for paging on SLO-related symptoms (for example, elevated error rates and latency). (Improved visibility into service behavior, faster incident detection and debugging, and a monitoring approach that scaled with a growing number of services.)
- DigitalOcean: Provide customers with managed Kubernetes monitoring using Prometheus-compatible metrics. — DigitalOcean integrated Prometheus-style metrics collection in its Kubernetes ecosystem and commonly pairs it with Grafana dashboards to visualize cluster and application metrics. (Simplified operational monitoring for Kubernetes workloads and improved observability for both platform teams and end users.)
Provider Equivalents
- AWS: Amazon Managed Service for Prometheus (AMP)
- Azure: Azure Monitor managed service for Prometheus
- GCP: Google Cloud Managed Service for Prometheus
Frequently Asked Questions
- What's the difference between Prometheus and Grafana?
- Prometheus collects and stores metrics and lets you query them with PromQL. Grafana is a visualization tool that builds dashboards and charts from data sources like Prometheus. In practice, Prometheus is the metrics backend, and Grafana is the UI for exploring and presenting those metrics.
- When should I use Prometheus?
- Use Prometheus when you need reliable, near-real-time monitoring of applications and infrastructure using numeric time-series metrics (CPU, request rate, latency, error rate). It’s especially strong for Kubernetes and microservices, where services can be discovered dynamically and you want alerting based on symptoms (for example, high p95 latency) rather than only host-level checks.
- How much does Prometheus cost?
- Prometheus itself is free and open source, but you pay for the infrastructure to run it (VMs or Kubernetes resources), storage for time-series data, and operational effort (upgrades, scaling, backups). Managed offerings (like Amazon Managed Service for Prometheus, Azure Monitor managed service for Prometheus, and Google Cloud Managed Service for Prometheus) charge based on metrics ingested, stored, and queried, plus any associated data transfer and retention settings.
Category: monitoring
Difficulty: advanced
Related Terms
See Also