# What is APM (Application Performance Monitoring)?

> APM is the practice and tooling for monitoring application performance and availability - response times, error rates, throughput, and transaction traces - to detect and diagnose problems before users feel them.

Source: https://openobserve.ai/glossary/what-is-apm/
Published: 2026-07-08
Term: APM (Application Performance Monitoring)
Category: Tracing & APM
Related terms: what-is-distributed-tracing, what-is-observability, four-golden-signals, what-is-real-user-monitoring

---

**APM (Application Performance Monitoring)** is the discipline - and product category - of monitoring the performance, availability, and correctness of software applications. Where infrastructure monitoring asks "is the server healthy?", APM asks "is the *application* fast and correct for the requests it's serving?"

## What APM measures

- **Response time** - latency distributions (p50/p95/p99) per service and endpoint
- **Throughput** - requests or transactions per second
- **Error rate** - failed requests, exceptions, and their stack traces
- **Transaction traces** - per-request breakdowns showing time in code, database queries, and external calls
- **Dependencies** - service maps of what calls what, with health per edge
- **Runtime internals** - GC pauses, thread pools, connection pools, memory

These map directly onto the [four golden signals](/glossary/four-golden-signals/), with [distributed tracing](/glossary/what-is-distributed-tracing/) supplying the per-request detail.

## How APM evolved

Classic APM (New Relic, AppDynamics, Dynatrace era) worked through proprietary language agents that auto-instrumented your code and shipped data to the vendor. It was effective but created deep lock-in: instrumentation, data format, and pricing were all one vendor's. The modern stack replaces the proprietary agent with **OpenTelemetry** auto-instrumentation and lets you choose any OTLP-compatible backend - turning APM from a product you buy into a capability you assemble. See the [top 10 APM tools](/blog/top-10-apm-tools/) comparison for the current landscape.

## APM vs observability

APM is application-centric and workflow-packaged; [observability](/glossary/what-is-observability/) is the broader capability spanning logs, metrics, traces, and frontend signals. In practice the categories have merged - what matters is whether your platform correlates application traces with the logs and infrastructure metrics around them.

## APM with OpenObserve

OpenObserve delivers APM on open standards: OpenTelemetry auto-instrumentation feeds [traces](/traces/), RED metrics, and service dashboards, correlated with [logs](/logs/) by trace ID - without per-host or per-user agent pricing. See [open-source APM tools](/blog/opensource-apm-tools/) for how it compares.