The auction of each type of vehicle on respective days through the web often experienced a high surge of traffic from bidders. This unmanageable traffic reduced the operating speed of the application software and the users were unable to bid in the auction. Weblanza was seeking for help to investigate and identify the reason behind this to find out a solution.
The application failed functioning after 5 to 10 minutes of duration when the number of real time users went beyond 500 even though the objective of Weblanza was to accommodate 5000 bidders using this application software simultaneously without any concerns.
The server dedicated for this application software, based on cloud technology, was levied higher rates due to lack of its optimization. Also, the stability of the server was compromised due to the increase in web traffic from bidders.
Tested initially for 500 users simultaneously. Proceeded testing with 1000 to 2000 users when the initial evaluation was found to be impeccable. The application started displaying errors 30 minutes after the ramped up strength of 2000 users started using it. The reasons for the challenges were identified using specific methods. The server statistics were monitored and the issues revealed were notified.
After the software developers from Weblanza rectified the issues, the product was re-examined. The number of users were increased up to 5000. Informed Weblanza regarding further issues identified on analyzing the server statistics. After the final rectification, the application was found to be working fine with good server stability for 5000 users. The response time for each user was also found to be acceptable for a maximum load of 5000 users.
Rigorous Performance Testing was conducted using Jmeter. Issues were reported and managed using Jira. Monitored the CPU Utilization, Network and Memory Usage during Performance Testing. Testing and reporting proceeded till the CPU Utilization was found to be below 20-25% and the memory and network related parameters were within acceptable limits.
The entire performance testing was to be done in the production environment, which was a challenge for us, as the large number of transactions to be done during the performance testing process should not damage the production setup.
Since the performance testing was to be carried out without interrupting the production users, it was to be conducted during non-working hours when no real time user was engaged with the production server. We used agile testing methodology. The different types of tests executed were load tests, stress tests and endurance tests.
The application is currently able to handle a load of 5000 real time users with an acceptable response time for each user.
The application is currently able to handle a load of 9500 Transactions Per Hour as per the workload modelling and scripting planned initially.
Even though it was a tight schedule for us to work with this project as we were given only 2 weeks of time to complete both Scripting and Testing, we were able to deliver it within the stipulated time frame. We completed the entire Test Scripting/Code Designing, Execution, Analysis and getting the issues fixed within 2 Weeks of time.