Following up on my Rise of the ZFS startups? post if ZFS never makes it across to Linux it may not matter all that much. Because we'll always have Butter. (Thanks Paul!)
ZFS as we all should know started at Sun. Btrfs (Butter FS) started at Oracle. ZFS is wrapped in Sun CDDL legalese. Btrfs is wrapped in the FSF GPL legalese.
That's a point in favour of Btrfs on Linux when it comes to acceptance before we even consider the technology.
ZFS is shipping today and you can have a lot more functionality than Btrfs provides right now (or in the near future) but Btrfs does have an aggressive development plan. The team is looking to ship a 1.0 release by the end of this year.
The Btrfs feature set is interesting and contains what you'd expect. Writeable snapshots, SSD optimisations, Object level mirrors and stripes, checksums, high speed offline fsck and there's even a conversion utility for Ext3 file systems.
There appears to be no built in volume management so I'm assuming, assuming mind you, that they'll look for the existing LVMs to manage that. Not a bad design decision since all your favourite volume manager need to do is add support for Btrfs.
But now we get to the parts which don't exist at the moment.
The only RAID levels currently available are 0, 1 and 10. No RAID 5. No RAID 6. Steve Todd always said RAID 5 is hard to do correctly so I don't see those coming in the very near future. It appears to do wacky (Not in the fun way) things with NFS and it's not capable of hot sparing drives.
All of these add up to knock it out of the game if you're a storage startup who'd prefer to focus on writing value added software instead of spending your time on the undifferentiated heavy lifting.
Right now Btrfs is clearly not the best choice for the current generation of Storage Startups.
The next generation of storage startups?
Well that's a different question entirely.
