Does the number (or complexity) of calculations on an Analysis slow down a dashboard even if those calculations aren’t displayed, or used as filters on the dashboard itself?
Hi @dmarcus -
The short answer is no. Only calculated fields for dimensions, sorts, metrics, conditional formatting, and filter expressions are used, and can potentially slow down an Analysis/Dashboard. The increase in metadata for the calculated fields has a negligible effect on performance.
Rob’s correct. Furthermore, if you actually ARE using those calculated fields in your visuals/filters/etc, then moving them into the data set prep screen can result in better performance specifically if you are using SPICE. Non-aggregate calculated fields you build in the data set get materialized in the SPICE data set (precomputed). This and some other nice tips are in this blog post around improving dashboard performance: Tips and tricks for high-performant dashboards in Amazon QuickSight | AWS Big Data Blog
1 Like