 |
 |
On Thu, Oct 26, 2017 at 2:48 PM, Josh Welker <[log in to unmask]> wrote: > Kyle, are all your Glacier/S3 assets backed up by a person, or is it > automated as part of an IR software package of some sort? > Both. This past year, we started using a semi-automated process for objects not in the repository -- there's an article in the most recent c4l journal describing it. The third figure in the article is a simplified depiction of an archival collection, though this also how datasets are handled. In the case of our Omeka installation which contains mostly simple images, the transfer is accomplished by script. In the case of ETDs, we push the assets, releases, and nonredacted copies of assets in zipped folders keyed on DOIs using a mostly manual process though we used automation for the retrospective work. In addition, Bepress provides full metadata and assets on S3 as a regular service using their own structure. The process is still new -- we have only been using it a few months. However, I believe it is solid and will handle technology transitions well as the method could easily be adapted for different platforms. We plan to move to Samvera which is being customized to automatically add DOIs as we create assets. It wasn't until this email that I realized those assets also need to be Glaciated. However, as our installation will be hosted on AWS and I'm presuming live assets will be stored on S3 so copying DOI-keyed versions to another bucket that automatically transfers to Glacier should be trivial. If you're looking for a simple way to protect assets and metadata, versioning (can be used for metadata as well as assets) and robust storage are your friends. kyle