Performance glossaryApplication performance monitoring (APM) tracks how an application behaves in production: response times, errors, and slow database queries. It finds problems before users report them.
Application performance monitoring (APM) watches a live application from the inside. An agent installed on the server, or a library added to the code, records each request: how long it took, which functions and database queries used that time, which external services it called, and whether it ended in an error. The APM tool turns all of that into dashboards, traces, and alerts.
On a WordPress or WooCommerce site, that might mean seeing that checkout is slow because a shipping plugin calls an external API on every page load, or that one request runs the same database query hundreds of times. Without APM, problems like these tend to surface as vague reports that the site "feels slow".
What APM measures
- Response time: how long the application takes to answer requests, shown as averages and percentiles for each URL or transaction.
- Throughput: how many requests the application handles per minute.
- Error rate: the share of requests that fail, with the stack traces behind them.
- Database and external calls: slow queries, repeated queries, and slow third-party APIs.
- Infrastructure: CPU, memory, and disk use on the servers or containers that run the code.
How it works
Most APM tools combine three techniques.
Metrics
Counters and timings collected continuously, such as requests per minute and response times. Metrics show that something changed, and when.
Tracing
A trace follows a single request through the application, and across services in a distributed system, showing how long each step took. Traces show where the time went.
Code profiling
Profilers sample what the code is doing to pinpoint the slow function, plugin, or query. Many tools add anomaly detection on top, flagging unusual slowdowns or error spikes without anyone having to set a threshold.
APM and other kinds of monitoring
APM looks at the server side. It is one of several complementary views:
- Uptime monitoring checks from outside that the site answers at all. See website monitoring.
- Synthetic monitoring, such as real-browser monitoring, loads pages and scripted journeys in a browser on a schedule.
- Real user monitoring (RUM) collects timings from actual visitors' browsers, which is how Core Web Vitals are measured in the field.
A page can be slow on the server, which APM shows, or slow in the browser because of heavy JavaScript and images, which APM mostly does not. Most sites benefit from both views.
APM for WordPress sites
WordPress runs on PHP, so any APM tool with a PHP agent can monitor it. New Relic and Datadog are common choices, and some managed WordPress hosts include an APM view in their dashboards. When we track down a slow WordPress site, the traces usually point to one of a handful of causes:
- a plugin running heavy database queries on every request;
- slow external HTTP calls made while a page is being built;
- bloated autoloaded options in the database;
- cron jobs or background tasks competing with visitor traffic.
For development and staging, the free Query Monitor plugin covers similar ground for a single request, listing database queries, hooks, HTTP calls, and PHP errors. It is a debugging tool rather than continuous monitoring, so it complements APM instead of replacing it.
Fixing what APM finds usually lowers server response time directly, and with it every loading metric that follows.