AIQ Load Generator Sizing Table
Recommended guidance
EstimatedUse load-generator sizing tables as planning guidance, not as a substitute for calibration. Actual capacity perdepends serveron scenario complexity, browser usage, protocol mix, data, network latency, host resources, and monitoring overhead.
What affects load generation
- Browser-based virtual users versus protocol/API virtual users.
- CPU, memory, disk, network, and operating system configuration.
- Think time, pacing, response time, retries, and error rate.
- Test Node count, geography, and application-under-test behavior.
- Logging, screenshots, video, and evidence retention.
Validation workflow
- Run a single-user smoke test before load calibration.
- Increase concurrency gradually and monitor both AIQ and application resources.
- Identify whether bottlenecks are in load generation or
VM (varies greatly with application complexity)Assumes E7 or newer Intel cores or EPYC AMD cores @3Ghz+.Refer tothetableapplication. - Document
below to determine Test Nodeachieved capacity andconfigurationassumptionsneeds:for
planning.CPUfutureCoresMinimum MemoryMinimumUX-Level BrowsersMax likelyUX-Level BrowsersHAR Protocol Level VUJMeter Protocol Level tests48 GB4810020612 GB61215030816 GB816250401224 GB1224500601632 GB16321000803264 GB3232250016064128 GB32645000320128256 GB326410000640
Using the latest Chrome browser, poor linear scalability is experienced after 30 browsers regardless of the size of the test node.
Example System Requirement Calculation
If a test specification requires UX tests to run 1000 virtual users, and each test node has an 8 Core CPU with 12 GB RAM, then 20 test nodes are required.
Calculated as:
Test Nodes Required = Total Virtual Users / Virtual Users per Test Node(from the above example) 20 Test Nodes = 1000 Virtual Users / 50 Virtual Users per Test Node
Last updated: JulyOctober 17,6, 20242026
If you have any feedback on this topic, please contact us at documentation@appvance.ai