Hi, hadoopers. I'm finalizing the asynchronous interface jclouds presents as an abstraction across cloud storage providers this week. This offers portability across S3, Azure, Rackspace, Mezeo, Atmos Online, and Nirvanix storage clouds. jclouds also supports non-blocking i/o and google-app-engine http engines, which may come in handy for performance or portability.
I noticed your native s3 project has a pretty clean design. I think it could be quite straight-forward to integrate this in a way that would work in jclouds. That said, there are always gaps and technical ownership is one way to ensure we get through them. Jclouds is looking for help on this, and the win for hadoop is achieving greater cloud portability. Is anyone interested in helping out and taking the lead on this? I got your back, and will facilitate the jcloudsy stuff as needed. Cheers, -Adrian founder jclouds http://code.google.com/p/jclouds
