Scaling 1 millions users Redline13
How to scale 1 million users in JMeter
ABOUT THE ClIENT
- The client requirement is to put 260 requests per sec which would be approx 1 million in an hour.
- The client provided us with different API Endpoints and the number of times the particular API should be executed per second as follows:
- We calculated the number of times the API should be executed per minute and per hour accordingly, as follows:
PROBLEM AND CHALLENGES FACED
- The client provided the Postman Collection.
- Headers are missing in some HTTP Requests in JMeter.
- We were having authentication issues.
- Each API request had to be executed with a different number of Users.
- When we pass a csv value in the request, the format of the value is automatically changed.
- As we have limited data, DB restore was needed after each execution.
- Executing a million APIs through JMeter is not preferable as JMeter is not that capable. This might require a large capable.
- When executing these APIs through JMeter, the CPU usage exceeded 80%.
- 5 CSV files were required for this execution, but RedLine13 allows only 1 file free for execution.
OUR SOLUTIONS
- We have converted the postman collection into JMX.
- We have added Headers in HTTP Request.
- We have a correlate authentication token using Bean Shell Assertion.
- We tried to use a concurrency thread group but somehow it did not work so we created a separate thread group for each and every API request.
- We have changed the CSV Format to a txt file.
- We said the dev team to clear the DB.
- We executed these APIs through RedLine13 using 8xlarge (32 vCPUs) machine.
- We executed these APIs through Redline13; CPU usage was highest at 5.2%, and an average of 44% for the Load generator.
- We buy a Basic Redline13 plan to execute 1 million requests within an hour.
TOOLS AND TECHNOLOGIES
- We used the JMeter tool for creating a script.
- We tested the execution of the script in Redline13.
- To more about Load and Performance Testing, click here.
BENEFITS TO THE CLIENT
- We’ve achieved an average response time below 1 second.
- Each API can handle the load of different sets of users.
- As per the requirement, we sent 258 requests/second for an hour of Test Time, which added up to around 1 million requests.
- We really recommend Reline13 to help us to scale this much of the user in a cost-effective way.
