Prompt
What should I use for APM on a microservices app?
Latest observation
Oct 29, 2026
OpenAI APIWeb search: off
For a microservices app, a good APM choice usually depends on how much observability you want, your stack, and whether you prefer SaaS or self-hosted. The short answer:
Best general-purpose picks
- Datadog APM — best all-around for most teams
- New Relic — strong full-stack observability, easy to use
- Dynatrace — very powerful, especially in larger/enterprise environments
- OpenTelemetry + backend of your choice — best if you want flexibility and portability
If you want the safest default
If you’re starting fresh and want something that “just works,” I’d usually recommend:
Datadog + OpenTelemetry instrumentation
- Great microservices tracing support
- Easy service maps and dependency views
- Logs, metrics, traces in one place
- Good Kubernetes support
- Broad language support
If you want to avoid vendor lock-in
Use:
OpenTelemetry
- Instrument your services once
- Send data to Datadog, Grafana Tempo, Jaeger, New Relic, etc.
- Good choice if you expect tooling to change later
If you’re on Kubernetes / cloud-native and budget-conscious
Consider:
- Grafana Cloud with Tempo + Prometheus + Loki
- Or self-hosted Jaeger/Tempo for traces
This works well, but usually takes more setup and tuning than a SaaS APM.
What matters most for microservices APM
Look for these features:
- Distributed tracing
- Critical for following requests across services
- Service maps
- Helps visualize dependencies and bottlenecks
- Logs-traces-metrics correlation
- Makes debugging much faster
- Sampling controls
- Important to manage cost and volume
- Kubernetes support
- Usually essential in microservices setups
- OpenTelemetry support
- Future-proofs instrumentation
- Alerting and SLOs
- Useful for reliability tracking
Quick recommendations by situation
- Small team, want easy setup: Datadog
- Enterprise, deep diagnostics: Dynatrace
- Need simple SaaS and good UX: New Relic
- Want portability / open standards: OpenTelemetry + Grafana stack
- Already using Elastic heavily: Elastic Observability can be viable
My practical recommendation
If you’re not sure, start with:
OpenTelemetry instrumentation + Datadog or Grafana Cloud as the backend
That gives you:
- vendor flexibility
- strong microservices tracing
- a path to scale
If you want, I can give you a best APM choice based on your stack (language, Kubernetes or not, cloud provider, budget).