Application Performance Monitoring (APM)¶

Application Performance Monitoring (APM) is the full-stack performance analysis solution built by Guance with distributed tracing at its core. It follows standardized protocols such as OpenTracing and deploys a unified collection agent at the host layer to enable platform-level correlation analysis of trace data, infrastructure metrics, and application logs, achieving end-to-end observability from code to resources.

Core architecture: It adopts a single-host, single-agent architecture. DataKit is deployed on each application server as the unified data collector.

Core Features¶

Service Observability¶

1. Service Map: Use the topology diagram to intuitively view the call relationships between services, with support for key metrics such as request count, error rate, average response time, P99/P95 response time, and maximum response time.

2. Service Details: View service upstream and downstream, service overview, and correlated logs and traces. It displays anomaly trend charts for metrics (service request response time, error request distribution) and logs (error log count).

3. Performance Metrics: Supports switching between Service/Endpoint views to view request volume, response time, and error status of services or endpoints, and locate performance issues through search and filtering.

Traces¶

1. Trace Explorer: Search, filter, and export trace data. The powerful search capabilities allow you to filter and view trace data from any time period, quickly surfacing abnormal traces.

2. Trace Details Page: Use tools such as Flame Graph, Span list, and Waterfall Chart to comprehensively analyze trace performance. Whether the calls are synchronous or asynchronous, you can clearly track the performance details of every trace.

Error Tracking¶

Error Tracking Explorer: Provides aggregated analysis and tracking capabilities for the various errors generated in distributed traces, with support for viewing the historical occurrence trend of a specific error type and its distribution across different services, endpoints, or instances.

Profiling¶

Profiling Explorer: Use deep profiling tools such as Flame Graph to visually analyze application runtime CPU usage, method execution time, and more.

Trace-correlated Profiling: Correlates application-layer performance bottlenecks (such as slow calls and time-consuming methods) with the resource consumption of the underlying infrastructure.

Analysis Dashboard¶

Analysis Dashboard: Aggregates the core analysis data of application performance, including trace statistics (Spans, request volume, and errors), correlated anomalies (error logs), deep performance analysis (response latency, call count, service request distribution, etc.), and resource-anomaly correlations.

Monitoring & Alerting¶

APM Metric Detection: Performs rule-based matching and filtering on trace performance data. Define specific detection conditions (such as response time exceeding a threshold or specific errors occurring) to identify and filter requests that match abnormal conditions from the full trace data.

Storage Billing¶

The system counts the number of trace_id values in the current Workspace and applies tiered pricing.

Specific billing rules and data storage policies (such as retention period) can be configured separately. See Data Storage Policy.

For more billing rules, see Billing Method.