...
Publish only what is needed. Not all of your data may need to be shared. Consider only those relative files to reproduce the results of your work or were necessary components that allowed you to come to your conclusions. See: Other data publishing considerations
Manage user expectations for downloading. Dataverse is not an active research notebook system. Users may need to download and analyze the datasets on a local system. Some of the most extensive datasets in our repository could take upwards of 60 hours to download under normal circumstances.
...
You don’t need to create a dataset for every file, and don’t confuse datasets and files. Datasets contain the metadata, the descriptive and administrative information about your project, and the files from the project. Files can be organized into a folder structure or tagged by keywords within a dataset. Do users need to understand the relationship of the file hierarchy, or do they need to filter them by keywords? Thinking about structure can make work easier for you and your users.
...
You can review our library tutorials, and read our guide that features an introduction to the DMPTool which can be used to help build your proposal and request feedback. See: Other data publishing considerations
Components of Responsible Data Management and Curation
...