Work Smarter, Not Harder: Discover Your OpenSearch Clusters Breaking...- Mike Oviedo & Rishabh Singh

Work Smarter, Not Harder: Discover Your OpenSearch Cluster’s Breaking Point With Redline Testing - Mike Oviedo & Rishabh Singh, Amazon Web Services Understanding how your OpenSearch cluster behaves under load is critical for capacity planning, stability, and regression detection. Traditional benchmarking applies static or sudden loads, making it difficult to determine the precise point a cluster’s performance begins to degrade. In this talk, we’ll introduce Redline Testing, a new feature in OpenSearch Benchmark that dynamically scales request load to find your cluster’s breaking point. Powered by a feedback-based client throttling system, this approach helps users identify this point and gives them the confidence that their cluster can handle sudden surges in traffic. We’ll dive into why sudden load spikes can lead to misleading results, how client ramp-up and real-time feedback coordination work under the hood, and how to configure and run your own redline tests to drive reliable capacity planning and performance validation. By the end of this session, you’ll know how to use this capability not only to prevent over provisioning and underperformance, but also catch regressions before they hit production. Whether you’re tuning queries or stress-testing a fresh deployment, this talk will help you benchmark smarter, not harder.