Openstack swift as deep storage


I would be happy to know what it takes in order to use openstack swift (object storage) as deep storage for druid ?

as far as I know, swift object storage is almost same as amazon s3, and have its own uri in the form of swift:///.


Generally, you need to implement the SegmentPusher and SegmentPuller interfaces to create a new deep storage. You can also choose to implement the SegmentKiller and one other that I’m not remembering the name of (on phone) for more robust functionality.

If you use Hadoop jobs to index your data, you also have to add the swift:// scheme to the IndexerGeneratorJob (there’s a switch/case statement in there with all the supported schemes), because of… reasons.

If swift has an S3 compatibility mode, you might also be able to use the S3 integration that way.