Sizing and Resource Recommendations for Kubernetes OpAMP Bridge

Review sizing guidance, resource requests and limits, Helm settings, and health probe recommendations for the Kubernetes OpAMP Bridge.

OpAMP Bridge runs as a Kubernetes deployment and enables remote management of OpenTelemetry Collectors running on Kubernetes. It uses the OpAMP protocol to manage the Collectors.

Sizing summary

  • Bridge replicas: A single replica of the Kubernetes OpAMP Bridge is sufficient to manage deployments scaling up to 5,000 OpenTelemetry Collector gateway pods.

  • Footprint: The Bridge has a lightweight resource footprint. CPU usage typically remains under 25m even under large rollouts.

Use the following resource requests and limits based on the number of managed OpenTelemetry Collector pods in your cluster:

Scale (Collector Pods) CPU Request CPU Limit Memory Request Memory Limit Observed Usage
Up to 3,000 pods 50m 200m 64Mi 128Mi ~11–22m CPU, 33–98 MiB RAM
3,000 to 5,000 pods 50m 200m 128Mi 256Mi ~15–20m CPU, ~128 MiB RAM

Example Helm configuration

Set the resource values in your values.yaml file for the splunk-otel-collector chart:

CODE 
opampBridge:
  enabled: true
  resources:
    requests:
      cpu: 50m
      memory: 128Mi
    limits:
      cpu: 200m
      memory: 256Mi

High-scale gateway health probe recommendations

When deploying large Gateway Collector workloads (>1,000 pods) managed by the OpAMP Bridge, mass configuration rollouts can cause temporary CPU spikes or probe contention on the API server.

To prevent unnecessary pod restarts during remote configuration updates, tune the Collector's liveness and readiness probes:

CODE 
agent: # or gateway: depending on deployment mode
  readinessProbe:
    httpGet:
      path: /
      port: 13133
    initialDelaySeconds: 15
    timeoutSeconds: 3
    periodSeconds: 10
    failureThreshold: 6
  livenessProbe:
    httpGet:
      path: /
      port: 13133
    initialDelaySeconds: 60
    timeoutSeconds: 3
    periodSeconds: 10
    failureThreshold: 12