Ingesting telemetry data into OpenSearch

Versions (relevant - OpenSearch/Dashboard/Server OS/Browser):
OpenSearch 3.x, Data Prepper 2.x

Describe the issue:
OpenSearch is a distributed search and analytics engine supporting various use cases, including observability, search, vector store, and security analytics. It integrates seamlessly with OpenSearch Dashboards, a user interface that enables users to discover and visualize OpenSearch data. This integration facilitates the creation of dashboards, alert monitors, anomaly detection, and post-mortem notebooks for root cause analysis.

Today, OpenSearch supports OpenTelemetry ecosystem through two primary ingestion methods:

OpenTelemetry Collector’s OpenSearch Exporter: Enables direct ingestion of logs and traces.
Native OTLP receivers in OpenSearch ingestion: Incorporates OTLP endpoints to ingest logs, traces, and metrics, supporting production-scale ingestion.

Request for Feedback

Your feedback will help us understand the path ahead for ingesting telemetry data into OpenSearch:

  1. Current Ingestion Strategy: What is your current approach for sending telemetry data into OpenSearch?
  2. Challenges Faced: What challenges or limitations do you encounter with either of the approaches (the exporter vs. native ingestion)?
  3. Preferred Method: Based on your current architecture, do you have a preferred ingestion method?

RFC:
This full RFC is open on the OpenSearch project GitHub repo under issue #18130:

It is also open under the OpenTelemetry project GitHub repo under issue #39707:

1 Like