Our client one of the world’s largest e-commerce marketplace for unique and creative goods, had it’s .com website and most of its related services hosted in self-managed data centers since 2005. In order to focus on strategic new features and services, they decided to migrate their IT infrastructure to Google Cloud Platform (GCP).

Softcrylic approached our client problems with a technology agnostic lens. We have deep experience in AWS, Azure as well as Google cloud and know how to effectively utilize those platforms in the most efficient and productive manner for our clients. Our client engaged Softcrylic as a partner in their cloud journey because of expertise in load and performance testing. For this engagement, we were the partner required to assure GCP was tested thoroughly to effectively run our client’s large e-commerce application.

Our client’s IT infrastructure supports more than 33 million users generating 1.2 billion events per day. Their .com web site receives 4 million page views and 40 million requests per hour from various users across the globe. When our client decided to migrate their IT infrastructure to the cloud, they wanted to make sure their cloud infrastructure can handle user traffic way more than the above-mentioned load. Also, our client had a very tight deadline to migrate their IT infrastructure to Google cloud before the US holiday season around Nov’18.

Our client was looking for ways to simulate a large user traffic load to their cloud infrastructure from across the world to ensure optimal performance. They were looking for a vendor who can perform such a large scale highly distributed load and stress testing, generating simulated user load from various locations around the world.

Softcrylic’s Test Engineering Practice provides load and performance testing as a service using open source and cloud services. We have a long history and deep knowledge in designing, developing and conducting large scale highly distributed load tests for our e-commerce clients. Based on our experience and the ability to execute large scale load tests from across the world, our client chose us as the prime vendor for their load and performance testing needs.

Softcrylic assembled a team of 3 performance test engineers and used open source tools and public cloud services such as AWS, Google and Azure to design, develop and execute large scale load tests. The engagement was well planned with discovery, design, development, execution, analysis and reporting phases. Our performance testing team collaborated with various technology teams at our client on this engagement.

Our first challenge in the project was to capture our client’s current user traffic and determine how much load we have to generate from our load testing tools for their normal and holiday traffic as there were no previous performance baselines. We used our client’s internal analytics tools to capture the user traffic over a period of time and come up with the appropriate load that we had to generate from our tests. Also, we had to conduct several low scale load tests during low traffic hours on their datacenter production environment, to establish baselines for key web pages, critical user actions, and the end to end transactions.

Once we had all our baseline and the numbers on how much load we had to generate, we designed, developed and conducted several large-scale distributed load tests. Our tests were distributed based on several test scenarios, user agents (desktop and mobile browsers) and geographical locations across the world.

Apart from conducting load tests, our team designed and developed a reporting solution to capture the performance metrics for every test and compare them to see the progressive improvements of the tests. Our team was also involved in analyzing, reporting and recommending performance optimization to our client’s team on a regular basis.

The softcrylic team was able to finish the load testing before the targeted migration planned date. Our client was able to migrate their web infrastructure and all related services successfully to Google cloud before Nov’18 holiday season.