How Tape à l’œil protects digital performance during retail peaks

Why they needed Gatling

Sales surges, API complexity, and mobile memory bottlenecks demanded more than intuition: TAO needed simulation-based certainty.

• Past traffic spikes had caused outages during key events
• Non-functional requirements had to be validated under peak concurrency
• A mobile app architecture issue caused RAM spikes during login
• Complex promotional tools required realistic load to verify performance

Gatling helped the team shift from reactive troubleshooting to structured, scenario-based testing that replicates real customer behavior.

The challenge: From theory to validation

TAO faced growing pressure to validate performance, not just assume it. One critical case involved mobile app users triggering 8+ API calls on launch. The app worked in QA, but when real users hit the platform at scale, RAM usage spiked.

Another challenge came from the way TAO handled pre-sale carts. Thousands of customers would prepare baskets before big sales events, only to check out en masse at 8 AM on launch day. Without simulating that burst load—including the cart transfer and promotion logic—the backend couldn’t cope.

Even with auto-scaling enabled, their environments couldn’t be trusted blindly. “Autoscaling isn’t magic,” Nordine explained. “We needed to warm up our pods before the real load hit.”

What they achieved with Gatling Enterprise

• Reduced memory usage by consolidating multiple mobile API calls into one
• Validated performance improvements after switching JSON serializers
• Simulated complex carts of 10+ items for sales-day benchmarking
• Ran warm-up tests to trigger autoscaling before traffic surges
• Built dedicated test scripts for mobile and web customer journeys
• Migrated from Scala to Java for easier scripting and broader adoption
• Performed load testing exclusively in staging while mimicking production behavior

Solution: simulate what matters, not just what’s easy

The team built specific load testing scenarios using Gatling Enterprise:

• A mobile script mimicking full user behavior: login, browse, cart, checkout
• A web simulation using closed injection models for homepage, product lists, and cart events
• Load tests for backend services like the commercial operations engine to validate complex pricing logic
• Performance regressions tracked using Gatling dashboards and correlated with Dynatrace and Datadog

What Tape à l'oeil Says

“Before Gatling, going live felt like crossing our fingers and hoping it held. Now we can simulate, measure, and move forward with clarity.”

Nordine El Mojahid, Head of Digital IT

The result: Fewer surprises, more confidence

• Improved memory usage by optimizing startup calls in the mobile app
• Detected and fixed issues in promotional pricing engines before campaign days
• Validated architectural changes (e.g., switching JSON libraries) through load simulation
• Reduced backend load by consolidating multiple endpoints into a single optimized one
• Logged 15–20 performance-related tickets annually, many directly linked to Gatling testing

What’s next for Tape à l'oeil

• Automate tests into CI/CD via GitHub Actions
• Run monthly performance simulations to catch regressions before they hit production
• Adopt reference-tier benchmarking for performance thresholds
• Correlate Gatling results with Datadog metrics using upcoming integrations
• Explore AI-powered reporting and test suggestions via Gatling Studio
• Use Postman-to-Gatling imports to simplify test script creation

‍