AWS-based automated performance testing solution to handle loads, identify bottlenecks, ensure SLAs, and deliver reliable performance for the new architecture.
Client
Leading health coaching organization with a global network
Goal
To ensure automated performance testing for SLAs using a cloud-based system with near 100% reliability on new architecture
Tools and Technologies
JMeter, Datadog, AWS API Management, Dockers orchestrated with Kubernetes, Azure CDN
Business Challenge
The client’s LMS and certification applications, built on a microservices-based architecture, required robust performance testing. Existing solutions struggled with encryption of APIs, varying network bandwidths, multi-region responses, and other critical parameters.
The goal was to create a cloud-based performance testing framework that could meet predefined SLAs, eliminate bottlenecks, ensure cost efficiency, and deliver near 100% reliability.
Solution
- Defined performance goals and identified bottlenecks
- Developed performance scripts using JMeter
- Ran bulk user load tests from three load generator machines using JMeter distributed load model
- Validated transactions across downstream applications through:
- Load tests with an incremental and iterative approach
- Various network simulation techniques
- Monitored load tests using Datadog APM
- Developed a need-based spin-up of performance environment
Outcomes
- Created 47 API test scripts using complex custom bean shell sampler and post processer
- Executed 30+ test cycles covering load, stress, scalability, endurance, spikes
- Identified nine critical performance bottlenecks including:
Middleware (API connector)
Database performance (mainly with indexes and orphan views)
Automated shift left performance testing and integrated DevOps
- Ensured cost efficiency with optimized AWS ecosystem
Enhanced overall system reliability from 73% to near 100
