Performance testing terms, without vendor fog.

Taurus is an open-source automation-friendly test runner format that can orchestrate tools such as JMeter and other performance engines.

BYOC means bringing your own compute or private location so runners execute near systems that should not be public.

Run artifacts are the files a performance run needs or produces, such as scenario bundles, logs, result exports, manifests, and reports.

Virtual users are simulated users or clients that execute a test scenario concurrently to create controlled load against an authorized target.

RUM — passive measurement of real browser sessions via a small JS beacon that reports page-views, Core Web Vitals, JS errors, and slow XHR calls.

Synthetic monitor — a continuous probe that fires periodic checks (HTTP, API workflow, browser flow, DNS, SSL, …) from one or more regions and alerts when the target fails.

Core Web Vitals — Google's standard set of user-experience metrics: Largest Contentful Paint (LCP), Interaction to Next Paint (INP), and Cumulative Layout Shift (CLS).

p95 and p99 latency show the slow tail of response times: 95% or 99% of requests were at or below that value.

RPS measures how many requests a system or test sends each second; it is useful only when paired with latency, errors, and request mix.

Error budget — the slice of allowed unavailability under an SLO (e.g. 99.9 % over 30 days = ~43 minutes of allowed downtime).

SLO — a target percentage (uptime, latency p95, error rate) measured over a rolling or calendar window, with an associated error budget.

Uptime — the percentage of time a service is responding correctly to its consumers, typically measured by a continuous synthetic monitor.