Knowledge Centre / Performance Centre
Testing methodology

NVMe Storage Testing for Hosting

Measure sequential and random I/O while explaining why latency matters.

Testing principles

  • Document hardware, software version and configuration.
  • Use the same workload for every test.
  • Run multiple passes and report averages and outliers.
  • Separate synthetic benchmarks from real server workloads.
  • Publish limitations and avoid guarantees based on one result.

Metrics to capture

Record CPU utilisation and clock behaviour, memory use, disk latency, network latency, error logs and the game-specific health metric such as TPS, MSPT, server FPS or save duration.

Suggested report format

  1. Objective
  2. Test environment
  3. Workload
  4. Results
  5. Limitations
  6. Conclusion
Replace example measurements with real 24 Play test data before publishing performance claims.