TRAY boosts enterprise POS system efficiency and reliability

Challenges

Before adopting Gatling Enterprise, the TRAY QA team faced several significant challenges in performance testing:

Time-taking endeavor to mimic production setup scenarios: Before using Gatling, TRAY found it time-consuming to replicate their geographically dispersed production environment, which involves multiple venues, locations, and brands. With Gatling, they can now accurately mimic their production setup with ease.

Performance challenges with a growing number of clients: As TRAY started serving larger clients with thousands of users across thousands of stores, they wanted to ensure consistent performance despite their growing client base. As TRAY continued to expand, performance became a major area of focus due to the increasing volume of transactions they handled each day.

Manual and time-consuming processes: Before Gatling Enterprise, performance testing was a manual and time-consuming process, requiring up to three days to collect and analyze data.

Solution

TRAY’s primary use case involves simulating the foundational flow of a staff member taking an order, authorizing a credit card, and completing the transaction. This process involves strenuously testing their user journeys with a concurrency of up to 20,000 users.

Load testing: TRAY performs load tests to ensure their systems can handle peak user loads. For example, their "place order" flow was optimized significantly, reducing response times from 18-20 seconds to just 2 seconds.

Continuous improvement: By integrating Gatling Enterprise into sprint cycles, they continuously monitor and improve API performance. This iterative approach has led to substantial performance gains, with some APIs showing up to 200% improvement in response times.

Improved performance testing framework: Gatling's test-as-code approach, programming concepts and reusable components streamlined TRAY’s testing process, allowing them to write and run tests more efficiently.

Enhanced reporting and data analysis: The detailed graphs and data tables provided by Gatling Enterprise made it easier to analyze performance metrics quickly. This reduced their data processing time from two days to just three hours.

Scalability and auto-scaling capabilities: Gatling helped TRAY implement effective auto-scaling strategies, ensuring that their systems could handle high user loads without unnecessary resource expenditure.

Results

Place-Order and 40-Column-Report are TRAY’s most used scenarios. They run these scenarios/simulations for 200 users, with the goal of generating up to 20,000 concurrent requests. The numbers provided are the average response times of the scenarios/simulations. The decreasing time indicates an improvement in their response time.

Enhanced user experience: The optimization of their systems has led to smoother and faster transactions, enhancing both customer satisfaction and employee efficiency.

Scalability and cost efficiency: Gatling's performance testing has enabled them to manage their resources better, avoiding unnecessary auto-scaling and reducing operational costs.

Utilizing chain builders with Gatling, what took days now takes hours. They aim to streamline their Gatling output to decrease this to 15 minutes. The detail and ease of access to the graphs and reports allows the TRAY team to monitor performance, identify areas for improvement and share data internally beyond just the QA team.

Gatling Enterprise has revolutionized their performance testing processes, providing TRAY with the tools to enhance their systems' efficiency, scalability, and reliability. The improvements they’ve seen since adopting Gatling have been substantial, positioning them well for future growth and success.

Future plans

Looking ahead, TRAY aims to integrate Gatling Enterprise fully into their CI/CD pipeline using CircleCI, reducing test execution times to 15 minutes and ensuring continuous performance monitoring.

They also plan to expand their testing scenarios to cover more APIs and further improve their system's resilience and scalability. At present, TRAY runs 30 simulations every two weeks, moving forward they are looking to have over 100 simulations run through Gatling Enterprise regularly to ensure performance, stability and scalability.

‍