OpenSearchCon Europe 2025

Every OpenSearch workload behaves and evolves differently. Users strive to understand their production workload’s behavior by running performance tests. But this is often done with generic data and basic queries, which can inaccurately represent their use case. The best way to determine its behavior is by benchmarking with real-life data and customer queries. To do this, users will need to recreate their actual workload, which is difficult as there are many factors to consider. Although it seems daunting, recreating production workloads and their behaviors do not require scientific methodologies and formulas but rather techniques and intuition.

In this interactive session, we’ll use a feature in OpenSearch Benchmark to convert real customer data into a sharable custom workload that recreates patterns and behaviors seen in production environments. We’ll also show how that custom workload can be used to save time, discover optimizations, and lower costs.

After the session, attendees will have developed intuition on workload dynamics, gained practical skills they can apply to their own environments, and understand that recreating workload behavior is an art, not a science.