Performance Engineering


PERFORMANCE TESTING


We use JMeter for performance testing

First, we analyse the application architecture either 2-tire or 3-tire.

Analyse the math of application as below format.

We deliver Performance Testing Services for the variety of applications including:

Web Enterprise Applications

Service Layer

Client-Server Applications

With many of the tools including Load Runner, JMeter, RPT.

Performance testing case study using JMeter Challenge:

We had a non-profit client having a member portal which interacts with many of the applications at the enterprise level, which also includes the Service Layer and External bodies, the challenge was that when Login happens, it use to take more than 30 seconds for login while the SLA as required was only about 5 Seconds.

How did we resolve?

Our Approach:

  • Understanding the Architecture of the Application.
  • Differentiate the layers for Enterprise Application.
  • Figure out the Response time taken for one user in isolation at each layer to see if we can meet the SLA with the existing design of the system, this includes exercising the response times even on the external servers.
  • Service Layer Response Times Identified.
  • GUI Layer Response Times Identified.
  • Accessed the server logs of each layer, and separate applications to know the pattern of usage of each of the component and their response times.
  • Came up with the WLAM as required by understanding the patterns of the user.
  • Got the WLAM reviewed, with an approval.
  • Designed the Scenarios in JMeter, including creating Tests for both GUI Layer and Service Layers separately.
  • Included the logic controllers to call service layer scripts in sequence and as required with the Web Layer Scripts.
  • Customized the Scripts by parameterizing and manual correlation as required.
  • Pilot run for 5 users to ensure the correctness of the scripts by including necessary assertions and checks at the database layers to ensure the record creation in the backend.
  • Used Dynatrace at the Enterprise level on each layer of the application, wherever we have access to it.
  • For external Servers, we made sure that the transactions as sent are received and responded with in the SLA as defined for the external layers.
  • Tagged the requests as required to ensure that we looked at the correct traffic to measure the response time.
  • Ran the script in combination to identify the timelines taken at the each layer, and identified the bottlenecks which included connection pooling issues, along with some code tuning as required.
  • Implemented the solutions across one after another.
  • Ensured the SLA met for less load.
  • Ran the scenario with Load combination as required.
  • Ensured that we met SLAs and certified the customer for improved performance.

Value additions: We also covered sceanrios including cluster server issues, exercised the sceanrios in absence of few clusters and ensured that the load is balanced well, in stress scenarios we ensured that right messages were displayed to the user, also that we made few components down by desing and ensured that the rejections do happen at the necessary application layers and appropriate messages are displayed to the end users.

Happy Customer towards the end with an improved performance in for portal, which allows millions of customers to access the same.