Scrape Prometheus metrics with the OTel Collector
- Latest Dynatrace
- How-to guide
- 1-min read
- Published Jun 23, 2026
The OTel Collector can scrape Prometheus endpoints and forward the resulting metrics to Dynatrace. Choose one of two deployment modes based on your scaling and reliability needs.
Standard (recommended)
A metric ingestion architecture that auto-scales, provides redundancy, and handles high-volume Prometheus workloads on Kubernetes. Recommended for most production environments.
Simplified
A single Collector for a small and static set of endpoints that don't need auto-scaling or redundancy.
How to choose
| Criteria | Standard | Simplified |
|---|---|---|
Best for | Production environments | Small and static target sets, demos |
Architecture | Target Allocator, scraper, and gateway pools | A single Collector |
Auto-scaling | Yes, with Horizontal Pod Autoscalers | No |
Redundancy | Yes | No |
Target discovery | Automatic via | Simple service discovery or static endpoint configuration |
Platform | Kubernetes | Any |
Scaling
If the simplified deployment runs out of headroom but you don't yet need the standard architecture, you can run multiple independent simplified Collectors, each scraping a distinct set of targets (for example, one per namespace or team).
- This keeps the same single-Collector configuration.
- You must partition targets yourself.
- There is no shared auto-scaling.
Because of this, we recommend the standard setup (which scales automatically) for most use cases.
For general information about scaling the Collector, see How to scale the OTel Collector.