AIP support storage through the file system and, via the s3 sub-project, in Amazon S3. You can use any other kind of persistent storage by implementing a few classes.
You must create a class that extends the Artifact trait. The only required methods are url, which must uniquely idenfity an individual dataset, and exists which should return true only if the data exists in its entirety. Add other methods to your class as necessary to support reading and writing to storage.
You must create a class that instantiates your Artifact class from a URL. See org.allenai.pipeline.s3.CreateCoreArtifacts as an example. Then you can create a mix-in trait that Pipeline classes can mix in to be able to support your storage class. See org.allenai.pipeline.s3.S3Pipeline for an example.
You must implement an ArtifactIo class that can read/write objects to your storage, using the methods in your Artifact class.
With these classes in place, you can specify a URL corresponding to your storage in Pipeline.rootOutputUrl to use your storage for persistence of any Producers persisted via a Pipeline.persist() method.