Laravel monitoring glossary

Application Performance Monitoring — what it captures, how it differs from logging and observability.

The three pillars (metrics, logs, traces) and how observability differs from monitoring.

In-process vs out-of-process vs PHP extension — how each architecture affects request-path overhead.

Latency percentiles explained, why averages lie, and how the gap between p95 and p99 exposes tail problems.

The 0 to 1 application performance index — formula, T threshold, and why percentiles often beat it.

The most common Laravel performance bug, explained with Eloquent examples and the eager-loading fix.

How error trackers group repeated exceptions into issues using class, stack frames, and normalized messages.

Why Telescope isn't production-ready beyond tiny apps, mitigations if you must, and what to use instead.

The difference between with() and on-demand relationship loading, and how each relates to the N+1 trap.

What distributed tracing is, when a Laravel monolith doesn't need it, and when it's worth the setup cost.

What OpenTelemetry is, how to set it up in Laravel, and when to pick it over vendor SDKs.

A trace is the full record of one request's journey — tree of spans sharing a trace ID, rendered as a waterfall.

One unit of work inside a trace — a DB query, HTTP call, or function with start/end times, attributes, and parent links.

Two strategies for sampling traces without losing signal. Head is cheap; tail is smarter.

A unique ID tying every log, span, and downstream request together. How to add one to Laravel.

Latency, traffic, errors, saturation — Google SRE's minimum monitoring signals applied to Laravel.

The difference between indicators, internal objectives, and external agreements. Plus error budgets.