TL;DR

• Use dashboard filters to reduce data volume being loaded all at once (i,e. Date range, Category, User Group)


• Dashboards will load slow if they contain a lot of widgets; for best performance, keep widget count < 15 when creating dashboards


• Dashboards with custom fields and advanced features such as merged results and pivot will take longer to load


• Large number of data attributes/columns will affect dashboard loading time

The most important element of dashboard performance is the underlying SQL query performance. Each widget on the dashboard runs a SQL query that takes time to execute on the underlying database. Below are some considerations for creating performant dashboards or widgets:

• Data volume has the greatest impact on performance. The more data that is returned in a widget on the dashboard, the more memory resources will be consumed. Widgets returned with many thousands of data points will use more memory. Utilize filters to limit size of data or rows of data values. A dashboard can also be modified to run on filtered values by default, please see this article for details.


• Limit the number of widgets on a dashboard. Running queries for each widget will consume resources which in turn will increase loading time (time to fully load all widgets) for a dashboard.


• Advanced features, such as merged results, custom fields, and table calculations, consume memory. The more post-query processing features used, the more memory is consumed and as a result, slow down dashboard load. 


• Pivoted dimensions consume memory. The more dimensions that are pivoted in a widget, the more memory is consumed when the dashboard is loaded. If the dimension you are pivoting has high cardinality (many unique values), there will be a column for each value. You could limit the displayed column count, or utilize filters to select the dimension values that you're most interested in, as opposed to showing everything at once.


• Having many columns and rows consumes more memory. Consider limiting columns or breaking up the dataset to multiple widgets/dashboards. 

*Widget: A single saved visualization that can be created from the data explorer to understand and analyze data. Netskope Library also offers a  number of out-of-the-box widgets that can be used to build dashboards.

*Dashboard: Dashboards are made up of one or more widgets to help users get a quick view of related content.