Slide 5 of 7

Getting instrumented

Two paths to Level 2

PathBest forHow it works
OpenTelemetry SDKAny environment — K8s, VMs, bare metalAuto-instrumentation libraries added to your app
Instrumentation Hub (public preview)Kubernetes — zero-code fast pathGrafana Beyla observes at the OS level, no code changes

What is Beyla? A lightweight agent that observes your applications at the operating system level, capturing traces without any code changes. It outputs standard OpenTelemetry data. Note that Beyla is newer technology — evaluate it carefully before using in production.

Problems solved

ProblemSolution
Instrumenting services takes months of dev workOTel auto-instrumentation libraries cover most frameworks automatically
Don’t know which services lack instrumentationBoth paths surface instrumentation gaps.
Vendor lock-in with proprietary agentsOpenTelemetry is the industry standard.
On Kubernetes and want zero-code instrumentationInstrumentation Hub + Beyla (public preview)

The tradeoff

OpenTelemetry SDKInstrumentation Hub
Works anywhereKubernetes only
Proven, widely adoptedNewer, still maturing
Requires SDK additionNo code changes
Industry standardPublic preview

Both paths lead to the same destination

Traces flowing → RED metrics → Service visibility → Level 2 achieved

Script

Application Observability and Knowledge Graph are powerful, but they need trace data to work. How do you get your services instrumented?

There are two paths. The proven path is the OpenTelemetry SDK. You add auto-instrumentation libraries to your application — they capture traces for common frameworks automatically: HTTP servers, database clients, message queues. This works everywhere: Kubernetes, VMs, bare metal, any language. The vast majority of teams use this approach, and it’s the industry standard.

If you’re on Kubernetes and want to move fast without touching code, Instrumentation Hub is worth knowing about. It’s currently in public preview. Behind the scenes it uses Grafana Beyla, a lightweight agent that observes your application at the OS level and captures traces without code changes. It’s newer and still maturing, so evaluate it carefully for production use.

Both paths get you to the same destination: traces flowing to Grafana Cloud, RED metrics for your services, and the visibility you need for Level 2.

/
Previous Next