TelemetryBackendPlugin

Purpose: Pluggable OTLP backends for traces, metrics, and logs Location: packages/floe-core/src/floe_core/plugins/telemetry.py Entry Point: floe.telemetry_backends ADR: ADR-0035: Observability Plugin Interface

TelemetryBackendPlugin separates telemetry collection (OpenTelemetry SDK, enforced) from backend storage/visualization (pluggable). This enables organizations to use existing observability infrastructure while maintaining standardized telemetry emission.

Note: OpenTelemetry instrumentation is enforced across all floe components. This plugin only controls where telemetry data is sent.

Interface Definition

The live ABC is TelemetryBackendPlugin in packages/floe-core/src/floe_core/plugins/telemetry.py. The snippet below is a conceptual excerpt for the public contract.

from abc import ABC, abstractmethod
from typing import Any

class TelemetryBackendPlugin(ABC):
    """Plugin interface for OTLP telemetry backends.

    Configure backends for storing and visualizing OpenTelemetry
    traces, metrics, and logs.
    """

    name: str
    version: str
    floe_api_version: str

    @abstractmethod
    def get_otlp_exporter_config(self) -> dict[str, Any]:
        """Generate OTLP Collector exporter configuration.

        Returns:
            Dictionary containing OTLP exporter config for the backend.
            Example for Datadog:
            {
                "exporters": {
                    "datadog": {
                        "api": {"key": "${env:DD_API_KEY}"},
                        "site": "datadoghq.com"
                    }
                }
            }
        """
        pass

    @abstractmethod
    def get_helm_values_override(self) -> dict[str, Any]:
        """Generate Helm values for deploying backend services.

        Returns:
            Dictionary with Helm chart values for backend services.
            Example for Jaeger:
            {
                "jaeger": {
                    "enabled": true,
                    "collector": {"service": {"type": "ClusterIP"}},
                    "query": {"service": {"type": "LoadBalancer"}}
                }
            }
        """
        pass

Reference Implementations

Plugin	Description	Self-Hosted
JaegerTelemetryPlugin	Local/self-hosted observability reference path	Yes
DatadogTelemetryPlugin	SaaS APM and distributed tracing	No
GrafanaCloudTelemetryPlugin	Managed Grafana + Tempo + Loki	No

Related Documents

• ADR-0035: Observability Plugin Interface
• Plugin Architecture
• LineageBackendPlugin - Companion observability plugin