CDK autogenerates resource names for every resource in your stack. However some, like DynamoDB, can take a table_name argument, making it a statically named resource. There are a lot of use cases this can benefit, one I’ve used it for was for a table that needed to be accessed by both dev and prod environments, and so needed to be recreated (using from_table_attributes instead of from_table_name so that we have access to any streams) for non-prod environments for access.