...
a set of digital representations, intended for a Web audience, of physical objects, which range from museum objects of all descriptions to books to taped audio interviews to VHS tapes. (Go to https://digital.sciencehistory.org/catalog and try limiting For a good idea of the range of materials, go to our search results and limit your search by genre, format, or medium, to get a sense of how broad the range is.) In the description below, when we talk about “original files” we are talking about these digital representations, which take the form of computer files. We store the original files in Amazon S3.
descriptions of the files above, which allow us to find them, keep them in order, search them, and describe them to the public. We store the descriptions in a PostGreSQL database hosted and managed by Heroku.
...
We store the backups for the original files in a separate S3 bucket that automatically mirrors the contents of the originals. On a nightly basis, these are copied to a local server, and our IT staff is responsible for making regular copies of these backups to a local disk, and then storing a series of tape copies of them offsite on tape.
We Heroku offers a service allowing us to “roll back” or revert the database to its state at any point in the past four days. In addition, we store nightly database backups for the database in a dedicated S3 bucket. Our IT staff also makes nightly copies of this the s3 bucket to local disk. From there, it joins backups of the original files in offsite tape storage.
...