Wild guess, perhaps the consume on close behavior allows reuse of HTTP persistent connection? This is probably not the right behavior for large blob downloads, although might have worked for small compute RPCs. a268309c94e8da21635950fef5ebc8737641f4b6 introduced this class.
@adriancole any ideas? --- Reply to this email directly or view it on GitHub: https://github.com/jclouds/jclouds/pull/243#issuecomment-31794682
