Navigating the Open‑Source Distributed Tracing Landscape: Tools, Features, and How to Choose

Getting started with distributed tracing can be daunting because the field is filled with overlapping terminology, frameworks, and tools that can easily confuse beginners. This article provides an overview and classification of the most popular open‑source tools to help you grasp the overall landscape.

Distributed tracing connects information from different work units—processes or hosts—into a single event chain, allowing you to understand the flow of an HTTP request across many microservices in modern applications.

Most tools can be roughly divided into instrumentation libraries, tracers, analysis tools (backend + UI), or any combination of these. The distinction between instrumentation and tracer is not always clear, and the term “analysis tool” can be broad because some solutions focus on exploring trace data while others provide a full observability platform.

This guide lists only open‑source projects, though several commercial solutions such as AWS X‑Ray, Datadog, Google Stackdriver, Instana, and LightStep are also worth watching.

Apache SkyWalking

Originally created in 2015 as a training project for understanding distributed systems, SkyWalking grew popular in China and aims to be a complete Application Performance Monitoring (APM) platform. It focuses on automated instrumentation via agents and integrates with existing tracers like Zipkin and Jaeger, as well as service‑mesh components. SkyWalking is now a top‑level Apache project.

Apache (Incubating) Zipkin

Developed by Twitter and open‑sourced in 2012, Zipkin is one of the most mature open‑source tracing systems and inspired many modern tools. It provides a full tracing solution—including instrumentation libraries, a tracer, and an analysis UI. Its B3 propagation format has become the de‑facto standard, and it is natively supported by Envoy Proxy and other tracing solutions.

Haystack

Created by Expedia, Haystack offers APM‑style capabilities such as anomaly detection and trend visualization. It relies on OpenTracing for instrumentation and can ingest data in other formats via components like Pitchfork.

Jaeger

Originating at Uber and open‑sourced in 2017, Jaeger moved to the Cloud Native Computing Foundation (CNCF). It draws inspiration from Google’s Dapper and Zipkin but has evolved beyond them. Jaeger uses the OpenTracing API for instrumentation and provides a lightweight analysis UI suitable for development and highly resilient environments, and it is the default tracer for Istio.

OpenCensus

Google’s open‑source project that serves as both a tracer and an instrumentation library. Its tracer can export data to open‑source analysis tools like Jaeger, Zipkin, and Haystack, as well as to vendor solutions such as Instana and Google Stackdriver. An OpenCensus agent can act as an out‑of‑process exporter, allowing applications to remain agnostic of the backend analysis tool.

OpenTracing

OpenTracing, hosted by the CNCF, provides a near‑standard API for instrumentation. It offers many language‑specific libraries (e.g., JAX‑RS, Spring Boot, JDBC). Several tracers—including Jaeger, Haystack, and commercial vendors like Instana, LightStep, Datadog, and New Relic—implement the OpenTracing API, and Zipkin also offers compatible implementations.

OpenTracing + OpenCensus

The two projects have announced a merger, with a provisional roadmap and code proposals outlining the direction of the combined effort.

Pinpoint

Developed by Naver in 2012 and open‑sourced in 2015, Pinpoint provides APM features such as network topology, JVM telemetry, and trace views. Instrumentation is performed via agents that can be extended with plugins, allowing instrumentation without code changes. It supports PHP and JVM applications and offers extensive framework and library coverage.

Veneur

Initiated by Stripe, Veneur acts as an observability data pipeline focusing on spans. It includes local “sink” agents that receive spans, extract or aggregate data, and forward it to external systems like Kafka. Veneur defines its own span‑focused data format (SSF), allowing metrics to be embedded in spans or aggregated from them.

Dapper

Dapper, described in a 2010 Google paper, is the conceptual ancestor of many tools listed here (Zipkin, Jaeger, Haystack, OpenTracing, OpenCensus). Although not a downloadable solution, the paper remains a key reference for understanding the primitives and design decisions behind modern distributed tracing.

W3C Trace Context

The W3C Distributed Tracing Working Group publishes the Trace Context recommendation to address interoperability challenges among different tracers by standardizing propagation formats.

Overview of All Projects

The tools differ in which components they provide: some offer only analysis (e.g., SkyWalking), some provide full stacks (e.g., Zipkin, Pinpoint), and others focus on specific roles such as instrumentation or tracing.

Original source: "A guide to the open source distributed tracing landscape" (https://www.infoq.cn/article/f5bBD119Zv49aak_eByE)