Data should be released as early as possible. If they belong to a publication, they should ideally be released when the corresponding manuscript is released. This enables the referencing and linking of the dataset in the publication to take place. Sometimes however there are reasons why the data should not be immediately revealed, such as protection of intellectual copyright. In this case, publication should be considered at a later time. For this reason, many repositories allow an embargo period to be set. During this time, the data are not accessible but are automatically opened afterwards.

If the dataset continues to grow after the initial publication, it can be updated. Many repositories allow versioning to take place. This ensures that different versions are accessible with their own unique DOI, whereas the top level DOI that is linked in a publication also remains valid. For datasets that are collected over a longer period of time and/or for which prompt publication is important, regular updating or even updating of the datasets in real time (“dynamic data”) is ideal. There are however not many repositories that support this.