Load tests for Black Friday 2021

For many online retailers, "Black Friday" represents the annual high-point both in per-day takings and simultaneous customer visits, and it is critical that the virtual shops run by such businesses can withstand this resultant high load. For the peak season in 2021, one of our longest standing clients, an online fashion retailer, expected an even greater increase in their order volume compared with the previous year, as well as likely the highest customer numbers to date. In addition, the substantial changes to the online shop infrastructure during the year were not backed up by significant experience of running the system for resilience in its new configuration. To assess and fortify the webshop in preparation for the peak season, a task force was formed out of members of development and operations teams, and here TNG played a leading role.

Working alongside other departments and external contractors, developers subjected the production setup to nightly cloud-based load tests that allowed a thorough assessment of the health of the system. Realistic test scenarios were reproduced in the form of JMeter test plans, simulating customer journeys through the shop, while the system could be monitored using both Elastic Stack, Promtheus, and Grafana, as well as in-house analysis tools. All relevant performance problems could be iteratively identified and fixed.

That this was a success could be seen on Black Friday weekend, when technical problems and performance disruptions were not only so minimal as to go unnoticed by customers, but were avoided entirely. The 2021 diet of load tests also afforded many insights into the webshop's performance and let to a round of suggested measures by way of preparation for yet further growth in the ensuing year.