Overview of SkyWalking UI: Dashboard, Topology, Tracing, Profiling, and Alerts

SkyWalking UI Overview

SkyWalking is an application performance monitoring tool for distributed systems, designed for microservices, cloud‑native and container‑based architectures such as Docker, Kubernetes, and Mesos.

The SkyWalking UI provides default visualizations for observing distributed clusters and consists of several areas: function selection, reload control, and time selector with language toggle.

Dashboard

The dashboard includes four categories: APM, Database, Self‑Observability, and Web Browser, each displaying metrics at Global, Service, Instance, Endpoint, and Database dimensions.

APM – Global

Throughput ranking (CPM)

Service response time ranking (ms)

Unhealthy service ranking (Apdex)

Endpoint response time ranking (ms)

Response time percentiles p99, p95, p90, p75, p50 (ms)

Response time heatmap (ms)

Concepts: Service: a set of workloads providing the same behavior. Instance: an individual workload within a service. Endpoint: request path or method signature for a service.

APM – Service

Service Apdex

Average response time (ms)

Success rate

Throughput (CPM)

Apdex curve

Percentiles p99 … p50 (ms)

Success rate curve

Throughput curve (CPM)

Endpoint throughput and response time rankings, success rate ranking, etc.

APM – Instance

Instance throughput (CPM)

Success rate

Average response time (ms)

JVM CPU usage

JVM memory usage

JVM GC time and count

APM – Endpoint

Endpoint throughput ranking (CPM)

Average response time ranking (ms)

Success rate ranking

Throughput and response time curves

Percentiles p99 … p50 (ms)

Success rate curve

Database

Average response time (ms)

Success rate

Throughput (CPM)

Percentiles p99 … p50 (ms)

Slow query list (ms)

Overall throughput and success rate rankings

Topology

The topology view shows service relationships, with selectors for individual services or custom groups, and allows inspection of metrics such as type, Apdex, success rate, response time, throughput, and percentiles for services and links.

Trace

The trace page enables searching by service, instance, endpoint, trace ID, and time range, and displays segment lists, service lists, span lists with timing and hierarchy, and three trace visualizations: list, tree, and table.

Concepts: Trace: a complete distributed call chain. Span: a timed logical unit within a trace. Segment: a collection of spans belonging to the same endpoint.

Common errors caused by code exceptions or network failures can be pinpointed by examining span details.

Profiling

Performance profiling captures periodic thread‑stack snapshots, aggregates them, and estimates method execution time based on consecutive identical stacks.

Create Profiling Task

Select target service.

Enter endpoint name (first segment operation name).

Choose monitoring start time.

Set monitoring duration window.

Configure monitoring interval to filter fast responses.

Set maximum sample count to limit data volume.

Only one profiling task can run per service at a time.

Analysis Result

After completion, segments appear on the right; selecting a segment shows detailed timing, and the “Analyze” button reveals method‑level details such as class, method, line number, and duration, highlighting the slowest stack.

Profiling Advantages

Precise problem location down to code line.

No need for repeated instrumentation, reducing development effort.

Minimal impact on target and monitoring systems.

On‑demand usage with controllable overhead.

Alarm

The alarm page lists all triggered alerts with filter options for service, instance, endpoint, and their relationships.

Note: This article uses SkyWalking version 8.2.0; UI may differ in other versions.