Fleek IT Solutions > Load Testing > Chalets Booking – Mobile App Load Testing

Chalets Booking – Mobile App Load Testing

Chalets booking app is an Arabic mobile app having great features to find chalets, filtering, pictures, pricing and booking options etc. Client had a very bad experience with performance of the app when they started their first marketing campaign, performance issues was a big set-back for their business. According to problem statement, load balancers and auto-scaling was in place, but application and infrastructure was not able to handle the load during marketing campaign.

Client contacted Fleek IT team in order to do a load and stress testing of the mobile app and find out performance bottlenecks in application, infrastructure and database. The biggest goal was to make sure that app performance is in a good shape before our next marketing campaign.

After understanding the complete requirements and business goals, Fleek IT team proposed an cost-effective and reliable solutions using following technical stack

  • Jmeter – an open source tool designed to load test functional behavior and measure performance
  • Jmeter plugins – a custom set of plugins for Jmeter to generate additional reports and, provide flexibility in scripting to generate more realistic load and many more features
  • Charles proxy – used as proxy to capture the requests and API calls
  • AWS EC2 instance – Load generation server
  • Blazemeter or redline13 as a load generation cloud

As a best practice we always keep an eye on potential issues during scripting, planning and of-course during test executions. Our experts identified few common issues quickly and notified developers:

  • Unwanted and duplicate API calls were made for many transactions
  • No server side or client side caching/local storage was implemented
  • Load balancer was not working as expected, auto-scaling configurations was not defined correctly and was causing errors to users during auto-scaling. However we recommended to first optimize application based upon single instance and see the bottleneck. This approach will help in planning infrastructure and related costing.
  • Chalet images was not compressed
  • Database queries was not optimized and multiple database functions were used in where conditions which was causing very high CPU utilization.