Skip to main content

Product Introduction

Overview​

APM looks at request health, call paths, errors, and goal attainment from the application and service point of view. Agents report applications, services, and instances into a directory. Engineers then search traces, endpoints, and errors by time, service, and environment, and receive alerts when error rate, latency, throughput, or traffic absence crosses a line.

Infrastructure metrics stay in Monitoring. APM answers which hop was slow, which service failed, and whether a downstream dependency dragged the call down. Redis, PostgreSQL, and similar dependencies that have no agent can still appear on the service topology when they show up in a call.

Note: The product UI and APIs provide onboarding, observation, and alerting. The data-plane deployment status is Beta. Agents connect only to a trusted in-region endpoint. Public-token ingestion is not supported.

Core Advantages​

  1. Install by language, then services appear in the directory
    Create an application, then follow the offline instructions for Java, Python, Node.js, Go, or .NET. The instructions point at the in-product download and the selected region's internal endpoint.

  2. From a health summary to one call
    The home page shows health, SLOs, alerts, and top metrics. The service pages show whether availability or latency goals are met. Explore searches traces, endpoints, and errors. Sensitive attributes are masked.

  3. Goals and alerts stay on the service
    Set an availability or latency SLO on a service, and choose an evaluation window. Alert policies use only error rate, P95, P99, throughput, and no-traffic. The service option already includes the environment, and an organization is required. An alert keeps the trend and evidence from the moment it fired. Closing it does not let later live data rewrite that history.

  4. Uses platform organization, notification, and Alert Center
    Data is limited to the current organization. Notification channels come from System Management. Regions come from Node Management. A policy can notify people and copy an event to Alert Center. Alert Center does not write that copy back.

Use Cases​

  • Check a release for slower or failing services
    Open the abnormal service, then use traces and grouped errors to see whether this service failed or a downstream dependency slowed down.

  • Hold availability and latency goals
    Configure SLOs for core services and read pass or fail on the home page and the service page.

  • Hand application failures to the on-call rotation
    Alert on error rate, latency, throughput, or missing traffic, with the evidence from that evaluation, and optionally copy the event to Alert Center.