The Btrfs file-system has various "shiny" features like support for copy-on-write snapshots, cloning, transactions, sub-volumes, SSD optimizations, transparent LZO/Zlib compression, and many other advanced features by Linux file-system standards. The problem with Btrfs is that the next-generation file-system is still in development and hasn't yet been proven via years of use and testing yet as being a stable and reliable solution. Most Linux distributions continue using the EXT4 file-system by default, but now there may be snapshot support coming back to EXT4.
Amir Goldstein at CTERA, a NAS device company, has been working on EXT4 snapshot support as they wish to have this feature for their network-attached storage units but aren't interested in yet deploying Btrfs for its copy-on-write snapshots. Amir has been working on this support for some time, and just yesterday sent out a fresh set of 36 patches that add this experimental support to the EXT4 file-system.
The set of 36 patches from yesterday can be found on the EXT4 mailing list
. The latest EXT4 snapshot patches are developed on GitHub
Obviously the work is too late to land for the Linux 3.0 kernel. It's possible that snapshots for EXT4 may be ready by the Linux 3.1 kernel, but there's still no 64-bit support and some of the EXT4 developers are concerned about this feature. Developers want to be certain that nothing regresses in terms of non-snapshot code-paths for EXT4. There's this message thread
with some discussion due to these patches touching some critical EXT4 code-paths.
In justifying the performance of EXT4 with snapshots, Amir even uses the Phoronix Test Suite
with Phoronix Global results he made
(from this message
). He should upgrade though to Phoronix Test Suite 3.0/3.2 in order to take advantage of OpenBenchmarking.org