Tangent is a global agency who create powerful digital products and experiences for some of the world’s largest brands and organisations, including Amazon, Regus, SAP and REED. As part of their client service level agreements (SLA), they guarantee to keep their client’s sites available, up and running 99.99% of the time, even during peak periods.
In early 2022, Tangent engaged Zoonou’s independent load testing services to optimise performance of their clients’ sites. Their client, one of the UK’s largest electricity distribution network operators, manages the power infrastructure for approximately 8 million households. It’s a priority for the client, in the event of extreme weather or a high-voltage fault, that its site remains operational for customers, so they can access vital information and services.
While they can’t predict the weather, Tangent wanted to provide their client with certainty that the site architecture was scalable and performant.
Automated test suites, designed to replicate real-world behaviour, were created by Zoonou’s Performance Test Analysts. The tests were executed using an in-house solution that extends the functionality of Apache JMeter and is powered by Amazon Web Services.
Over the course of 12 months, Zoonou executed more than 50 test runs across 4 periods of performance testing. Each round of testing focused on specific goals:
The load tests were monitored in real-time by Zoonou alongside the Tangent team. This allowed Tangent’s developers to implement changes to the site between tests and see the reshaping of results, run after run.
The data output from each round of performance testing enabled Tangent to make informed changes to the site’s architecture and content delivery network. The implementation of these changes in configuration ensured the most effective and cost-efficient support for varying user loads and scenarios. The results ensured Tangent met their SLA commitments as demonstrated in the graphs below.
In this graph, from the first load test, the site returned errors (orange line) throughout the test, indicating the site was unable to support increased traffic.
After 12 months of repeated testing, investigation and infrastructure updates a final test was executed. The graph demonstrates the site is now able to support a higher number of concurrent users, with no errors reported.