Skip to main content

APM

APM is the application performance monitoring app built into base14 Scout. It turns OpenTelemetry traces and metrics into service health views, dependency maps, grouped errors, database and messaging analysis, and links to individual traces.

Services table with throughput, P99 latency, and error rate sparklines for every instrumented service

Charts read pre-aggregated rollups and detail views read raw spans, and the two have different retention limits. See Raw trace limits for what that means when a chart has data and its occurrence list does not.

Tabs at a Glance​

TabWhat it answers
ServicesWhich services or server operations have abnormal traffic, latency, or failures?
Service MapWhat calls what, and which node or connection is unhealthy?
TracesWhat happened during an individual request?
ErrorsWhich exceptions are recurring, and what did their latest occurrences do?
Database OperationsWhich database statements or operations are slow, frequent, or failing?
Messaging QueuesWhich messaging operations are busy, slow, or failing?
InfraAre the hosts or Kubernetes nodes beneath the services healthy?

Services, Service Map, and Traces are always present. The other four depend on rollups and metrics an administrator enables per deployment.

Use Cases​

Triage a Latency Regression​

  1. Open Services and sort by P99 latency to find the service that moved.
  2. Open Service Detail and check its Spans table for the operation carrying the increase.
  3. Open the operation and pick a slow occurrence to see the full trace in traceX.
  4. If the trace spends its time in a query, continue in Database Operations.

Find Which Dependency Broke​

  1. Open Service Map and isolate the failing service.
  2. Inspect its unhealthy outgoing connections for call volume, latency, and error rate.
  3. Open the suspect dependency's Service Detail or its logs in logX.

Decide Whether It Is the Code or the Machine​

  1. Confirm the regression in Services.
  2. Open Infra and select the instances you know run that service, then check CPU, memory, network, and filesystem.
  3. If the resources are healthy, the cause is in the application path, so go back to the operation and its traces.

FAQ​

Which APM tab should I start with?​

Start with Services. It compares every instrumented service by throughput, P99 latency, and error rate over the same window, which is the fastest way to see which service moved. Every other tab is reached by drilling down from something you found there.

Why do some APM tabs not appear in my deployment?​

Errors, Database Operations, Messaging Queues, and Infra depend on rollups and metrics an administrator enables per deployment. Until the rollup exists, the tab stays hidden rather than rendering an empty page. Services, Service Map, and Traces are always present.

Does APM read the same data as traceX?​

Both read the spans your instrumentation sends, but at different resolutions. APM aggregates them into rollups so fleet-wide views stay fast, and traceX reads raw spans for one request. That is why an APM chart can cover a period its occurrence list cannot.

  • Getting Started - Scope, filters, and trace limits
  • Services - Service and operation RED metrics
  • Service Map - Dependencies and connection health
  • Errors - Grouped exceptions and stack traces
  • traceX - The full trace explorer
  • logX - Logs for the same services
  • k8X - The Kubernetes fleet underneath
  • RUM - How mobile clients experience these same services
Last updated on
Was this page helpful?