Performance

Load stress testing details and charts

Our system's performance primarily depends on the hardware, web server, and database server performance. Below are the details of our testing on the following server configuration. No special tuning has been applied to the web server and database.

CPU: Intel Core i7-6700K (Skylake-S)

RAM: 32 GB DDR4 SDRAM

SSD: Samsung EVO 870

Operating system: Ubuntu 24.04

Web server: Apache 2.4.58

Database server: MariaDB (MySQL) 10.11.7

Number of merchants: 5000

Number of terminals: 50 terminals per merchant

The stress testing was conducted by injecting pairs of requests: one to create a new dynamic bill from the terminal and another for payment notification for this new bill from the payment channel. The response time was measured within the application, starting from the moment control was passed from the web server to the PHP script until the final response was generated and handed back to the web server. Response time is measured in milliseconds. Additionally, up to 100 milliseconds should be added for processing inside the web server.

The stress test duration was approximately 1 hour, during which we managed to raise the request rate to up to 20 API requests per second (more than 1000 requests per minute) with no significant slowdown in processing time. The average processing time did not exceed 50 milliseconds, occasionally reaching up to 100 milliseconds.

STATS PER SECOND

Number of API requests:

Requests per second

Response time:

Response time

STATS PER MINUTE

Number of API requests:

Requests per minute

Response time:

Response time