Load Generation Limits and Guardrails

Review execution limits and infrastructure guardrails for the Load Generation feature in Scale Test.

Capacity and Execution Limits

Scope	Parameter	Limit	Notes
Per scale test run	Max scripts (tasks)	25	JMeter only. Scripts and tasks are synonymous.
Per scale test run	Max scripts (tasks)	10	Playwright only. Scripts and tasks are synonymous.
Across all runs	Max concurrent users	4000	Playwright only.
Across all concurrent runs per org	Max parallel tasks	100
Outside a booked scale test window	Max scripts (sanity runs)	5	Granted by exception. Applies to test runs outside the booked calendar window.
Outside a booked scale test window	Max Requests per Second (RPS)	20	For script validation only.

Book scale test days only in the same sandbox where you plan to test.
The parameters you submit when booking are final. Your actual test must match the booking inputs, including RPS and the maximum number of user logins per minute. Inputs that exceed these values are rejected.

Per-task RPS monitoring: Each task is monitored independently to prevent a single high-load task from affecting others running concurrently.
Per-task concurrent user monitoring: Concurrent users are capped at the task level, which is especially important for Playwright runs that use more memory per user.
Outbound IP allowlisting: Backend access requires an outbound IP allowlist configured at the workflow level before you run a test.
Endpoint security: Add external backends by providing an IP address or hostname through the admin UI. Restricting access to My Domain only is no longer supported.

JMeter handles thread-based API load. Guardrails focus on RPS thresholds and connection pooling.
Playwright runs one browser instance per virtual user. Concurrent user limits per task are enforced strictly to prevent out-of-memory errors.

Did this article solve your issue?

Let us know so we can improve!