Send Data

The OpenSearch Observability Stack ingests telemetry through a standards-based pipeline. Applications emit traces, metrics, and logs using OpenTelemetry, which are collected, processed, and routed to their respective storage backends.

This section covers every layer of the ingestion pipeline, from instrumenting your code to configuring the collectors and processors that deliver data to OpenSearch and Prometheus.

Architecture

The following diagram shows the end-to-end data flow from your applications to the storage and query layer:

flowchart LR
    A["Application<br/>(OTel SDK)"] -->|"OTLP gRPC :4317<br/>OTLP HTTP :4318"| B["OTel Collector"]
    B -->|"OTLP :21890"| C["Data Prepper"]
    B -->|"OTLP HTTP :9090"| D["Prometheus"]
    C --> E["OpenSearch"]
    D --> F["OpenSearch<br/>Dashboards"]
    E --> F

Traces and logs flow through the OTel Collector into Data Prepper, which processes and indexes them in OpenSearch. Metrics are forwarded from the OTel Collector to Prometheus via OTLP HTTP, where they are stored and queried by OpenSearch Dashboards.

Supported Protocols

Protocol	Port	Transport	Use Case
OTLP gRPC	4317	HTTP/2 + Protobuf	SDK default, highest throughput
OTLP HTTP	4318	HTTP/1.1 + Protobuf or JSON	Browser, serverless, firewall-restricted environments

Both endpoints accept traces, metrics, and logs. CORS is enabled on the HTTP endpoint for browser-based instrumentation.

Default Endpoints

Component	Port	Protocol	Description
OTel Collector (gRPC)	4317	OTLP gRPC	Primary telemetry ingestion
OTel Collector (HTTP)	4318	OTLP HTTP	HTTP telemetry ingestion
OTel Collector (metrics)	8888	Prometheus scrape	Collector self-monitoring
Data Prepper	21890	OTLP gRPC	Trace and log processing
Prometheus	9090	OTLP HTTP	Metrics storage

Sections

OpenTelemetry

Core instrumentation framework. Learn about the OTel Collector configuration, auto-instrumentation for zero-code setup, manual instrumentation for custom telemetry, and sampling strategies for controlling data volume.

Applications

Language-specific guides for instrumenting your services. Covers Python, Java, Node.js, Go, .NET, and browser applications, plus dedicated guidance for AI/LLM agent observability.

Data Pipeline

Configure the backend processing layer. Covers Data Prepper pipelines for trace and log processing, Prometheus for metrics storage, and index management in OpenSearch.

Infrastructure

Collect telemetry from your infrastructure. Covers host metrics, container monitoring, Kubernetes observability, and cloud provider integrations.

Quick Start

To start sending data from any application, set two environment variables and run with auto-instrumentation:

export OTEL_EXPORTER_OTLP_ENDPOINT="http://localhost:4317"
export OTEL_SERVICE_NAME="my-service"

Then follow the language-specific guide in the Applications section, or jump straight to Auto-Instrumentation for zero-code setup.

Related Links

• Get Started — Platform overview and sandbox setup
• Investigate — Query and explore ingested data
• APM — Application performance monitoring views