Did you know that the Socrata platform supports dataset backups? This feature allows you to keep an up-to-date version of your dataset on hand within the platform itself, in case of a failed update or any other change you would like to revert.
- What is a backup?
- How can I create a backup?
- How many backups can I have on a dataset?
- Are backups publicly accessible?
- How can I restore my dataset from a backup?
What is a backup?
A backup is an exportable CSV of your Socrata dataset, which you can take at any point in time. Backups are not discoverable in your domain's catalog, do not have assigned 4x4s, and do not have any associated metadata. These files allow you to easily revert changes to your data if you find you've made a mistake in a recent update, or even if you just want to track changes to your data over time. (Any time you are making a manual update in the platform UI, feel free to make a quick backup first for peace of mind!)
Backups are not available for derived views (e.g. charts, maps, filtered views), since those asset types are inherently dependent on underlying datasets.
How can I create a backup?
Backups are created both manually in the data table as well as automatically by Socrata when certain conditions are met.
- Navigate to the grid view of your dataset by clicking "View Data" on the dataset's primer page, or appending /data to the dataset URL.
- Click on the "More Views" tab, expand the "Dataset Snapshots" section and click "Back Up This Dataset".
- Your backup will be added to this and is identified by a timestamp. Clicking on the backup link under "Dataset Snapshots" will export your backup as a CSV.
Backups will also be created automatically by Socrata, any time your dataset goes through a publication cycle. A publication cycle only occurs on a dataset update when certain conditions are met. These conditions are:
- Adding or removing a column
- Changing a column datatype
- Changing the API field name for a column
- Setting or removing a row identifier
Please note that standard data only updates will not automatically create a dataset backup*.
*This refers to updates through the Socrata Dataset Management Experience. This behavior may differ based on your data update process.
How many backups can I have on a dataset?
You can keep a maximum of 3 backups on an individual dataset at any time. Backups do not expire and will not be deleted unless you already have 3, in which case the oldest backup will be deleted when you create a new one.
Are backups publicly accessible?
Backups are public or private based on the visibility of the dataset itself. Backups on public datasets will be publicly accessible under More Views > Dataset Snapshots, and can be exported by any user; backups on private datasets are private and can only be accessed and downloaded by authorized users.
How can I restore my dataset from a backup?
To restore your dataset from a previous backup, click on the backup under the "More Views" tab in the dataset's grid view, which will export your backup CSV. You can then create a new revision to upload the CSV as a replace.