Title

Exploring Mendix's Performance Limits: Optimization for 1000 Concurrent Users

Million rows in a single table, 1000 concurrent users, how to optimize the performance of a Mendix application?

"Low" Code with "High" Performance: Unveiling Mendix's High-Performance Optimization Plan!

Cover

Background

Recently, we conducted performance testing and optimization for a large enterprise-level OA application developed for a client, built on the Mendix platform, and ready for launch.

This application is targeted at the entire enterprise, primarily providing process approval functions, with approximately 1 million approval submissions per year. Over 8000 employees use this OA system, with over 800 users online simultaneously during peak hours (9:00 AM to 10:00 AM).

Before optimization, the system could not handle even 20 concurrent users. After optimization, the system can easily handle 800 concurrent users. Below is a comparison of some key metrics before and after optimization.

My Task Page Performance Comparison Before and After Optimization

My Task Page Performance Comparison Before and After Optimization

As shown in the above data, the optimization significantly improved the system's performance. Next, we will analyze how the optimization process was carried out.

The optimization process is mainly divided into three levels:

  1. First Level: Reducing User Requests. At this stage, we mainly optimized the communication logic between the front and back ends, removed unnecessary requests, and reduced the number of network requests from the front end to the back-end server.
  2. Second Level: Reducing Database Requests. At this stage, we reduced the number of requests from the application server to the database server. The main measure was to optimize the SQL statements sent by the application server, changing from SQL statements that request one record at a time to those that request multiple records at once.
  3. Third Level: Reducing Single Database Query Time. At this stage, we no longer optimized the application logic but optimized the SQL query statements while ensuring consistent return results to reduce the database query pressure. For example, establishing appropriate indexes and optimizing query conditions.

The following text will elaborate on the above optimization process in detail.

First Level: Reducing User Requests

At this stage, we mainly optimized the communication logic between the front and back ends, removed unnecessary requests, and reduced the number of network requests from the front end to the back-end server. As shown in the figure below, when a user successfully logs in and enters the homepage, they will enter the My Task list. In this list, each row represents a pending approval form, and each approval form is associated with multiple historical approval records.