CUSTOMER STORY

How TRAY load tests a restaurant POS at 20,000 concurrent requests

Date

Thursday, December 17, 2026

Time

6 PM CET / 12 PM EST

Duration

45 min, pre-recorded session

You'll learn:

  • How to model production as reusable test code: multiple venues, locations and brands, built from chain builders for take order, authorize card and complete transaction
  • How TRAY sizes its load: Place-Order and 40-Column-Report run with 200 users to generate up to 20,000 concurrent requests
  • How load testing in every sprint pays off: Place-Order from 18-20 seconds to 2 seconds, with some APIs up to 200% faster
  • How to scale without overspending: using load tests to tune AWS auto-scaling so high load doesn't mean unnecessary resource spend
  • Where Gatling fits in the toolchain: CircleCI, AWS and Git today, and what comes next with a 15-minute test execution goal
  • How reporting and AI cut analysis time: from two days to three hours per test cycle, and what the team still checks by hand after reading an AI result

‍

This is for you if:

  • Your load testing is still manual, or analysis takes days instead of hours
  • You need to reproduce a multi-location production setup in a test
  • You work on SaaS, POS or payments systems where a slow transaction is felt at once, by every customer

‍

Ideal for: QA leads, performance engineers and engineering managers

Speakers

Henry Darana

Henry Darana

Head of QA

Tray

Diego Salinas

Diego Salinas

Content Marketing Manager

Gatling

Join this session

Free. Live or on demand: register and we'll send you the recording.

ABOUT THIS SESSION

Every second a POS spends on "place order" is a second a guest waits at the counter. Across thousands of stores, that delay becomes a business risk.

‍

TRAY builds point-of-sale software for enterprise restaurant brands. Its core flow: a staff member takes an order, authorizes a credit card and completes the transaction. As larger clients brought thousands of users across thousands of stores, Place-Order averaged 18 to 20 seconds, and performance testing was manual, taking up to three days to collect and analyze data per cycle.

‍

In this session, recorded by the TRAY team, you'll hear how they got Place-Order down to 2 seconds with test as code, load tests in every sprint and Gatling Enterprise, and how they now use Gatling AI to cut analysis from days to hours.

‍

Your all-in-one load testing platform

Design complex tests, manage global infrastructure, and turn results into action on one powerful platform.

Need technical references and tutorials?

Minimal features, for local use only