Member Article

The Load Testing Cheat Sheet: Three steps to avoiding website disaster for businesses

If sites go down, profits go with them. When Currys PC World crashed due to a user surge on Black Friday in 2016, they lost out in sales revenue, and suffered a loss of reputation during one of the most intense retail periods of the year. When it comes to peak online shopping days like Black Friday, major chains are reporting ever-growing volume, so delays and failures in loading simply aren’t viable in a competitive marketplace.

The more it happens, the more damage is done to brand reputation. While unit and functional testing are important parts of site security and performance, companies must acknowledge the significant problems that only load testing can uncover. Here are three load testing basics that can avoid website disaster:

1. Test your site early and frequently

It’s not only retailers who need to worry; all companies with an online presence need to load test. A product launch or marketing release are just two events that will generate higher traffic than normal for any business; failing to make sure sites can handle a major user influx will cost them dearly.

Today even the largest enterprise companies are adopting agile development methodologies. Applications that were once updated a few times a year are now updated on a monthly or even weekly basis. A decrease in time between code releases requires an extensive automated testing program. Testing an application’s performance under real-world conditions should be a vital part of every deployment.

Keeping a consumer’s interest is tough; a recent study from Google revealed that just half a second extra loading time sees site traffic drop by 20%. Therefore frequent and rigorous load testing is a must for maintaining successful applications.

It’s vital to leave enough time between test and launch to fix any problems a load test uncovers. Even testing two weeks in advance of anticipated heavy site traffic is probably too late.

2. Get to know your data

Load testing by itself isn’t enough to maintain website performance. Knowing how to interpret the data that comes out of a test is essential. Special attention should be given to session duration, network throughput and application performance.

Across the whole test, session duration should be constant, meaning access isn’t taking longer for some users than others. Network throughput should increase directly in line with traffic with no struggling. And of course, there shouldn’t be any errors in application performance.

Results, and how they differ from these ideals, are easily represented on a graph. Companies can then decide which steps to take to rectify issues.

3. Run the three key load tests: Stress, concurrency and disaster recovery

Load testing finds the theoretical breaking point of applications by exposing them to large numbers of users. It also assesses system response to common user scenarios and failure conditions. Whilst there are plenty of ways to load test a site, three methods are widely used.

• ‘Stress’ (throwing data at the site and seeing if it can cope) finds the bottlenecks in a website, application, API, or supporting infrastructure.
• ‘Concurrency’ examines what happens at predicted levels of traffic and is commonly used by retailers.
• ‘Disaster Recovery’ reveals how your site recovers from failures.

Carrying out any and all of these options is achieved by writing a script and running it from an on premise solution or SaaS platform. It’s a good idea to have your starting point for amount of load on the low side.

By using all of these methods, resulting loss of traffic (and income), is easily avoidable. In routinely analysing online performance and pinpointing bottlenecks in your application or website, companies will be able to identify issues before they impact business in order to ensure an optimised and unmatched customer experience.

This was posted in Bdaily's Members' News section by Carmen Carey .