Apache SkyWalking is an open-source application performance monitoring (APM) system for distributed systems in cloud native architectures, built for teams that run microservices across Kubernetes and container-based environments.

What it is

Apache SkyWalking is an Apache Software Foundation project, written in Java and released under the Apache-2.0 licence, that provides monitoring, tracing and diagnosing capabilities for distributed systems. Its repository describes it as an APM system designed especially for microservices, cloud native and container-based architectures, and its topic list places it in application performance monitoring, distributed tracing, observability, service mesh and eBPF territory alongside standards such as OpenTelemetry and Prometheus. It covers the three telemetry signals that observability teams usually assemble from separate systems: distributed tracing, metrics and logs.

The concrete problem it solves is fragmentation across the observability stack. Metrics, traces and logs from mature ecosystems such as Zipkin, OpenTelemetry, Prometheus, Zabbix and Fluentd are supported and processed together, and SkyWalking's native meter format plus the widely known formats from OpenTelemetry, Telegraf and Zabbix all run through the same script pipeline. Rather than replacing each of those systems, it consolidates them, adding service topology analysis, service-centric observability and APIs dashboards on top of data that would otherwise sit in disconnected stores. Storage is handled by BanyanDB, an observability database created in 2022 that aims to ingest, analyze and store telemetry and observability data, and the project states that one SkyWalking cluster can collect and analyze 100+ billion telemetry data points.

Key capabilities

• End-to-end distributed tracing with service topology analysis, service-centric observability and APIs dashboards.
• Agents for Java, .NET Core, PHP, NodeJS, Golang, LUA, Rust, C++, client JavaScript and Python, under active development and maintenance.
• The Rover agent, an eBPF-powered monitor and profiler that watches Kubernetes deployments and diagnoses CPU and network performance.
• Consistent metrics aggregation across SkyWalking's native meter format and external formats from OpenTelemetry, Telegraf and Zabbix through one script pipeline.
• A log management pipeline supporting log formatting, metric extraction and various sampling policies at high performance.
• Alerting rules set at service, deployment and API level, with forwarding of alarms and all telemetry data to third parties.
• Machine learning and AI analysis of observability data, including HTTP URI pattern recognition and automatic metric baseline calculation for intelligent alerting.

Who uses it and how

• Platform and SRE teams running microservices on Kubernetes use the Rover eBPF agent to profile deployments and diagnose CPU and network issues without instrumenting each service.
• Teams already committed to OpenTelemetry, Prometheus, Zipkin, Zabbix or Fluentd keep those pipelines and add SkyWalking for topology, dashboards and alerting over the same data.
• Large installations rely on the stated scaling envelope of 100+ billion telemetry data points collected and analyzed from a single cluster.
• Full-stack product teams combine the client JavaScript agent with backend agents for Java, .NET Core, PHP, NodeJS, Golang, LUA, Rust or Python to trace requests end to end.
• Teams that need API-level alerting use URI pattern recognition and automatic metric baselines to detect anomalies rather than hand-writing thresholds.

Getting started

Releases are downloaded from the project's releases page at skywalking.apache.org/downloads/, or the project can be compiled following docs/en/guides/How-to-build.md. A preview deployment can be set up quickly through the SkyWalking showcase documentation, and a live demo with the native UI and Grafana, plus a 3D Infrastructure Map powered by Horizon UI, is published on the website.

How it compares

SkyWalking sits alongside the tools it ingests from: Zipkin, OpenTelemetry, Prometheus, Zabbix, Fluentd and Telegraf, plus Grafana as a visualization layer. Where those projects each own one signal or one pipeline stage, SkyWalking folds traces, metrics and logs into a single APM system with its own database, BanyanDB, and its own alerting and telemetry pipelines.

When to use it — and when not to

Adopters must operate a SkyWalking cluster plus observability storage such as BanyanDB, and the README does not document the rest of the production prerequisites, with sizing beyond the 100+ billion telemetry figure left to the official documentation. It is a poor fit for teams that want a managed service with no infrastructure of their own. The README is also thin on installation detail: quick-start commands live in separate documentation, and the main repository points readers to the downloads page and a build guide instead.