How does indexing work in Spice?

Hello! I am using a fairly large data source loaded into SPICE, and am investigating how performance could be improved.

How does indexing work in SPICE? Is there any way to add indexes or sort keys to commonly filtered columns? Is this done automatically? Will sorting my data before hand improve Spice performance?

SPICE is compressed columnar data store, each columns are stored in its own file. Filters are applied at the SPICE storage level resulting into smaller datasets for visualization. SPICE uses a proprietary mechanism for storing, indexing and retrieving data and does not require indexing from QuickSight authors/admins.

I would recommend to start with a dataset that meets your specific business use case requirement by including only required columns and use filters in data preparation editor as applicable to limit the number of rows. You can always add columns to your dataset as your business requirements expand. Avoid using a single dataset to meet all of your dashboarding needs, instead focus on purpose build datasets to meet specific dashboarding and analytics needs.

In case you are running into performance challenges, please work with your account manager or open a support ticket.

Thank you