Skip to main content

Number of Concurrent Users and Related Availability Tests

Test Environment Configuration

  • Microsoft Azure virtual cloud host, Cent OS 7.5, single node
  • CPU 2.30 GHz, 4 cores and 8 threads
  • Memory 64 GB, SSD
  • With cache enabled

System Throughput

  • The chart below shows that peak throughput is 180 requests/s, average request response time under high concurrency is less than 1s, and the request success rate is 100% under 800 concurrent requests.
  • According to Microsoft capacity planning methodology, system reliability above 99% can be ensured at any time, even during daily peak load, as long as concurrency is guaranteed for 5% of the total active users.
  • Using retail as an example, in actual store or personal report usage scenarios, the maximum concurrency for a single node is 50. Based on a single-card data request response time of less than 1s, the number of active users this configuration can support is estimated as: 50 ÷ 5% = 1000 users.

image.png

Notes

  • The test above is based on a cloud host, namely a virtual machine. If the actual environment uses physical machines, performance can improve significantly.
  • The test above is a standard test. In actual use, targeted optimization can be performed based on data characteristics and usage scenarios.
  • As the product continues to iterate, performance metrics will continue to improve. Actual performance shall prevail.
  • System capacity can be increased by improving the computing capability of Server nodes.