What it does
Apache SkyWalking is an open-source application performance monitoring and observability system for distributed services and cloud-native infrastructure. Its backend analyzes telemetry, a web UI helps teams explore it, and agents or probes send data from instrumented services and supported ecosystems. The project's current site also documents ways to inspect model traffic and AI-agent sessions, alongside service, Kubernetes, and infrastructure monitoring.
SkyWalking is a complete observability stack to operate, not just an agent that forwards telemetry. The backend receives and analyzes data, a storage layer persists it, and the UI queries the results. Its project site describes signals including topology, traces, metrics, logs, profiles, and alerts, with language agents and integrations for common telemetry ecosystems.
Prepare the backend before instrumenting services
The official setup guide asks operators to prepare a database before starting the backend and recommends BanyanDB. The distribution includes server scripts, configuration, libraries, and the UI. The OAP backend exposes gRPC and HTTP endpoints; agents and probes commonly send to gRPC port 11800, while the UI serves on port 8080 and queries the backend over HTTP. The runtime guide lists Java 11, 17, 21, or 25.
A reasonable first task is to install the backend and storage, then connect one test service using an agent or OpenTelemetry-compatible receiver. For example, if a request passes through several microservices and slows after a change, use the service topology and trace to locate the affected hop, then examine related metrics or logs. That illustrates how the documented signals fit together; no SkyWalking instance was deployed for this research.
Decide whether to operate it yourself
The project is Apache-licensed software. Teams choose storage, size the OAP backend, configure collection, and keep compatible agents or probes running. That flexibility is useful for organizations that want control over their telemetry stack, but it involves more setup and maintenance than a hosted monitoring account. The official pages reviewed here did not establish a vendor-hosted commercial price or service tier.
Strengths
- Brings service topology together with multiple signal types, including traces, metrics, logs, profiles, and alerting.
- Maintains language agents and probes and supports established telemetry ecosystems such as OpenTelemetry, Prometheus, Zipkin, and Zabbix.
- Uses a modular backend with configurable storage and query components rather than requiring one fixed storage provider.
Watch out for
- Self-managed deployments require a Java runtime, a prepared storage backend, and configuration of the OAP server, UI, and collection agents or probes.
Best for
- Investigating service dependencies and request paths across microservices or Kubernetes environments.
- Teams that want an Apache-licensed observability stack and can operate its backend and storage.
- Collecting and correlating traces, metrics, logs, and profiles in one self-managed system.
Sources
- Apache SkyWalking — Observability for cloud-native and AI (Oct 26, 2026)
- Backend setup | Apache SkyWalking (Oct 26, 2026)
- GitHub - apache/skywalking: APM, Application Performance Monitoring System · GitHub (Oct 26, 2026)
- skywalking/LICENSE at master · apache/skywalking · GitHub (Oct 26, 2026)
- Backend storage | Apache SkyWalking (Oct 26, 2026)