Very nice write-up. I saw a video not too long ago about coming developments in storage devices that would require a significant but possible re-working of the VFS an perhaps block layer of the Linux Kernel. And it looks like, from this talk, that may _have_ to happen in chunks as time progresses.
Object storage is interesting, but difficult without vendor buy-in of your platform. I know a few object storage systems work by emulating a file system, but actually using a translation mechanism between the actual on-device file system and the emulated file system. One such example would be the Data Robotics Drobo devices. I'd also be wary of future patent troubles in the object storage arena, since lots of vendors seem to do it behind the scenes.
It would be fantastic to see representatives of Linux organizations working closely with hardware manufacturers! If Linux environments got to shape future devices rather than just be compatible with them, I think we'd start to see a significant change of paradigm of other hardware vendors. And since Linux has a solid foot hold in the enterprise arena, it's as good a place as any to start!