Export to Your Observability Backend

Mermin exports Flow Traces via the OpenTelemetry Protocol (OTLP), compatible with any OTLP-enabled observability backend or collector.

What You Need

To receive Flow Traces from Mermin, you need one of:

OTLP-Enabled Collector: OpenTelemetry Collector that receives OTLP and forwards to your backend(s)
OTLP Data Platform: An observability platform with native OTLP ingestion

OpenTelemetry Collector (Recommended)

The OpenTelemetry Collector provides the most flexibility:

Receives OTLP from Mermin via gRPC or HTTP
Processes, batches, and transforms telemetry data
Exports to multiple backends simultaneously
Provides buffering and retry logic

Example Configuration: See Mermin with OpenTelemetry Collector for a complete setup with OpenTelemetry Collector, including Mermin configuration and collector pipeline.

Basic Mermin Configuration

export "traces" {
  otlp = {
    endpoint = "http://otel-collector:4317"  # Collector's OTLP gRPC endpoint
    protocol = "grpc" # Optional; Mermin defaults to "grpc".
  }
}

OTLP-Compatible Data Platforms

Flow Traces work with any platform supporting OTLP trace ingestion:

Elastic Stack

Elasticsearch with APM Server or OpenTelemetry Collector ingests OTLP traces.

Use Case: Full-text search, complex aggregations, APM integration, machine learning

How to Connect:

Point Mermin → OpenTelemetry Collector → Elasticsearch exporter
Or point Mermin → Elastic APM Server (OTLP endpoint)

Example: See docs/deployment/examples/netobserv-os-simple-svc/ for OpenSearch (Elastic-compatible) deployment

OpenSearch

Open-source alternative to Elasticsearch with native OTLP support via OpenTelemetry Collector.

Use Case: Open-source search and analytics, cost-effective storage

Examples:

docs/deployment/examples/netobserv-os-simple-svc/ - Basic OpenSearch setup
docs/deployment/examples/netobserv-os-simple-gke-gw/ - GKE deployment with Gateway API

Greptime Ingestion

Greptime is a database designed for high-cardinality time series data that supports OTLP ingestion.

export "traces" {
  otlp = {
    endpoint = "http://greptime-standalone-instance:4000/v1/otlp/v1/traces"
    protocol = "http_binary"

    headers = {
      "x-greptime-db-name"       = "public"
      "x-greptime-pipeline-name" = "greptime_trace_v1"
    }
  }
}

Example: docs/deployment/examples/greptime_simple_svc

Grafana Cloud, Datadog, New Relic, Honeycomb, etc

Most commercial observability platforms support OTLP ingestion.

How to Connect:

Obtain your platform's OTLP endpoint URL
Configure authentication (usually API key or bearer token)
Point Mermin to the endpoint with auth

export "traces" {
  otlp = {
    endpoint = "https://otlp.provider.com:4317"
    headers = {
      "authorization" = "Bearer ${API_TOKEN}"
    }
  }
}

Examples: Coming soon...

Flow Trace Data Model

Each Flow Trace is an OpenTelemetry span containing:

Span Attributes:

Network 5-tuple: source/dest IPs, ports, protocol
Bidirectional counters: bytes sent/received, packets sent/received
TCP state: flags (SYN, FIN, RST), connection state
Kubernetes metadata: pod, service, deployment, namespace, labels
Community ID for flow correlation across monitoring points

Resource Attributes:

Kubernetes cluster name
Node name
Mermin version

Query Flow Traces using native backend query languages (TraceQL, KQL, Lucene).

Testing with Stdout

For local development and testing, output Flow Traces to stdout instead of OTLP:

export "traces" {
  stdout = {
    format = "text_indent"  # Human-readable format
  }
}

View traces in Mermin logs:

kubectl logs -f -l app.kubernetes.io/name=mermin