Dears,
I’m working on planning monitoring to services hosted in an on-prem OpenShift cluster using Dynatrace, and I’d like to determine the best monitoring approach: Full-Stack Monitoring vs. Application-Only Monitoring.
Current Setup & Goals:
- Deployment: On-prem OpenShift cluster
- Dynatrace OneAgent: Currently considering build-time injection for application-only monitoring (without deploying OneAgent as a DaemonSet)
Monitoring Scope:
- Application-Only Mode: Tracing, custom metrics, logs, RUM (frontend monitoring)
- Full-Stack Mode (Alternative): Host & Kubernetes infrastructure monitoring in addition to application monitoring
Questions:
1- Which monitoring approach is better for OpenShift on-prem?
If we focus on application-level monitoring only, will we miss critical issues root cause? what is the expected Dynatrace Host Unit (HU) vs. Davis Data Unit (DDU) consumption?
If we enable full-stack monitoring, what is the expected Dynatrace Host Unit (HU) vs. Davis Data Unit (DDU) consumption impact?
2- What are the best strategies to minimize unnecessary span collection, excessive metrics, and log ingestion to reduce DDUs?
3- If we choose Application-Only Mode, how can we still monitor OpenShift API, node performance, and Kubernetes events effectively?
Real-World Insights:
4- Has anyone implemented Application-Only Monitoring for OpenShift on-prem?
5- What challenges did you face, and how did you optimize monitoring without full-stack visibility?
