A dataset is a collection of JSON documents that can be of different types and have references to each other. You can think of a dataset as a “database” where all of your content is stored, whereas the document‘s types would constitute “tables”. Using GROQ or GraphQL you can always query and join data across documents within a dataset, but not across them. Typical applications of datasets are:
- operate with different environments for testing, staging, and production
- localization and segmentation across all content types
- different purpose content, but with same user access and billing
Enterprise users with the custom access control feature can set custom access control rules for each dataset. Users on regular plans can define public or private visiblity mode for their datasets.
Datasets can be created and managed using the
sanity command-line tool, e.g. by running
sanity dataset create <name> or
sanity dataset list. To see all dataset-related subcommands, run
Datasets can also be created and deleted in the project's management console, under the "Datasets" tab.
A dataset name may have a minimum length of 1 character, and a maximum length of 20 characters. It may only contain the lowercase characters a-z, the numbers 0-9, the "-" (hyphen) and the "_" (underscore) characters. Also, a dataset name must begin with a letter or number (a-z or 0-9).
Enterprise customers can take advantage of additional functionality for managing datasets in Sanity's architecture.
You can initiate dataset copying directly in the cloud and create aliases to hot swap between datasets without changing the underlying code for your project.