Case Studies

Accelerated performance and reduced error rate to 0.07% for an online food delivery app through AI-driven load testing

Conquered on: 03 . 09 . 2021 Download case study as PDF

About the client

TmmmT is an online food delivery application to order food that provides instant delivery for food, vegetables & fruits.

The application should be able to handle the load of 2308 concurrent users at peak hours where maximum orders placed should be 98 per minute

The Challenge

Location – Unable to place orders due to location issues. After location setup from backend, we were able to browse the application and place orders

Data Dependency - Limited set of test users due to which data needs to be created before every round of execution.

Load Testing activity was performed on production that might impacted the production on high

Load Testing Execution time slot was a concern but was handled by doing the activity in odd hours

The Solution

  • Gathered understanding of TmmmT application from the customer for all the workflows to be tested for all the 3 mobile applications .
  • Created the test suite to be used for performance testing with different set of users.
  • Data preparation (for some workflows) before each test and data segregation for different workflows.
  • Performed sanity run to validate the scripts.
  • Baselining the execution with low volume of concurrent users.
  • Load testing with the increasing load to 511, 1027 & 2308 concurrent users.
  • Understanding bottlenecks and recommending the fix to the customer in phase 1
  • Sharing the test results and maximum orders placed at instant of time with the client
  • Errors and Observation were shared after each test execution performed

Services we offered

Performance Testing

Benefits

  1. Response time for most of the key business transactions that were within the defined SLA for all the 3 mobile apps. 
  2. Final load test could successfully ramp up 2308 concurrent users with 0.07% error rate
  3. Maximum 98 orders could be placed per minute with 2308 users load test

Key Outcomes

99

reduction in error rate