# What is a Telemetry Pipeline?

> A telemetry pipeline (or observability pipeline) collects, transforms, filters, enriches, and routes logs, metrics, and traces between sources and destinations - controlling cost and data quality in flight.

Source: https://openobserve.ai/glossary/what-is-telemetry-pipeline/
Published: 2026-07-08
Term: Telemetry Pipeline
Category: Observability Fundamentals
Related terms: what-is-telemetry, what-is-opentelemetry-collector, what-is-log-aggregation, what-is-otlp

---

A **telemetry pipeline** (also called an **observability pipeline**) is the processing layer that sits between the systems producing telemetry and the destinations consuming it. Instead of shipping raw logs, metrics, and traces straight to a backend, a pipeline lets you **filter, transform, enrich, redact, sample, and route** data while it is in flight.

## What a telemetry pipeline does

Typical pipeline stages:

- **Collect** - receive data from agents, SDKs, syslog, cloud services, or message queues
- **Parse & transform** - turn unstructured text into structured fields, normalize formats, convert units
- **Filter & sample** - drop health-check noise, debug logs, or apply trace sampling before paying to store them
- **Enrich** - add Kubernetes metadata, GeoIP, service ownership tags
- **Redact** - mask PII, secrets, and tokens for compliance before data leaves your boundary
- **Route** - send security events to a SIEM, everything to cheap object storage, and a filtered subset to real-time analytics

## Why pipelines matter

Two forces made pipelines a distinct product category. First, **cost**: with per-GB ingest pricing, dropping 40% of low-value logs before ingestion is the single fastest way to cut an observability bill. Second, **control**: compliance teams need PII scrubbed in flight, and platform teams want to switch backends without re-instrumenting every service - a pipeline decouples producers from consumers.

## Building one

The open-source path is the [OpenTelemetry Collector](/glossary/what-is-opentelemetry-collector/), whose receiver → processor → exporter chains are telemetry pipelines by construction. Alternatives include Vector, Fluent Bit, and Logstash, plus commercial routing products. The trade-off is operating another tier of infrastructure at ingest scale.

## Pipelines in OpenObserve

OpenObserve builds [pipelines](/pipelines/) into the platform: real-time and scheduled pipelines can parse, transform (using VRL functions), redact, and route streams between sources and destinations - including [forwarding to external systems](/blog/sending-data-from-openobserve-pipelines-to-splunk-using-splunk-http-event-collector/) - without deploying separate pipeline infrastructure. See the [setup guide](/blog/how-to-setup-pipelines-in-openobserve/) to build one in a few minutes.