5 Best Practices to Avoid a Site Meltdown

End-to-End Transaction Performance Analysis
Thorough web load testing will result in errors and bottlenecks — maybe even a full-scale crash — and that's a good thing! The objective is to identify issues with garbage collection, caching, database connectivity and much more. This step also allows for collecting a large amount of data. If tested correctly, this provides enormous insight into network, infrastructure and application performance.

Ongoing Simulated and Real User Monitoring
Anticipate change going into a website peak period, and keep testing and monitoring for its impact on performance. Maintain a consistent baseline of clean-room measurements against which to benchmark performance. These measurements provide the guidance to identify issues. Also, augment this monitoring with performance measurements collected from actual user visits. In this respect you have both simulated and real-user monitoring data to provide benchmarks of optimum performance.

Contingency Planning
Despite the best testing, no system is perfect. Problems happen, sometimes out of the control of a site owner. Make sure you've designed your site so that it can gracefully handle system failures. Even if one of your vendors or external systems goes down, you need to be able to function, even if impaired, so consumers stay on your site and continue to place orders. If you need to then make the site unavailable overnight, for example, it can be painful but necessary to fix what can't be fixed while live.

Prioritize readiness. Test early, test often. Anticipate challenges. It's the only way to prepare for a major online event. Check out the infographic below to learn more.

Aaron Rudger is senior marketing manager for web performance at Keynote, a provider of mobile and website testing and monitoring. Aaron can be reached at aaron.rudger@keynote.com.