There's Talk Again About Btrfs For Fedora
Phoronix: There's Talk Again About Btrfs For Fedora
In earlier Fedora Linux releases there was generally a desire with having Btrfs become the default file-system. It's generally proposed to make the next-generation Linux file-system the default in Fedora, but every time in the end the idea has been dropped. With Fedora 19 due in mid-2013, Btrfs for Fedora is again being talked about...
in most benchies i see on Phoronix, EXT4 runs circles around BTRFS in most of the tests (except when you set compression in BTRFS).
Dont see why it should be default.
Most linux noobs probably use ubuntu anyway (i am one too).
Fedora is the other side of the Ubuntu coin, though, if you are starting first time on Linux it is probably one of those two, and I don't see why you wouldn't use lzo or gzip compression in btrfs, you save space and get better disk throughput.
Originally Posted by mayankleoboy1
I run it default on my Arch boot, and snapshots are am amazing way to do system restores.
Btrfs is like Duke Nukem, poised to ship way later than anticipated, so late that it's not even funny.
Btrfs has been "shipping" for a long time and many of us have been use it for a long time already. This discussion is merely about whether it is ready to take on the role of a default filesystem for Fedora. Filesystems are like wine. A new filesystem written from scratch simply takes a long time to mature and we cannot really rush that process. Comparable filesystems have either had long gestation periods before it reached public eye (ZFS) or were extensions to existing mature codebases (Ext*).
Originally Posted by mark45
I'm myself using Btrfs for like 18 months anywhere except the / partition, iirc Ubuntu were contemplating making it the default one like 2 years ago, but since then the only progress is that Ubuntu/Canonical doesn't think it's ready, nor does it contribute anything, just as with Wayland - it's pretty much waiting for Red Hat and others to do the dirty job. Plus iirc there was a guy who refused to work publicly on the Btrfs fsck which made Btrfs evolve even more slowly than it could.
btrfs seems to be taking longer than ZFS did. Which is ironic, since ZFS was more innovative for its time than btrfs. I would have expected btrfs to be developed faster than ZFS, since the concepts that ZFS pioneered were well established by the time btrfs development started.
Originally Posted by RahulSundaram
ZFS development started in 2001, and it was stable and production ready in 4 or 5 years (depending on whether you count OpenSolaris build 27 in 2005 or Solaris 10 in 2006 as a stable release).
btrfs development started in 2007, and it looks like it would be optimistic to expect btrfs to be the default filesystem in Fedora in 2013, which is six years later. And Fedora is generally considered a cutting-edge distro. You'd need to have btrfs used as default in a conservative linux distro to have a close equivalent to ZFS in Solaris 10.
btrfs development is so slow and aimless that I wonder whether btrfs will ever be chosen as the default filesystem for a conservative linux distro.
btrfs isn't really just about performance, it's about capabilities. btrfs is hugely different from a simple partition format like ext4; btrfs incorporates volume management and redundancy and all sorts of other features that are usually layered on top of simple formats with tools like LVM and mdraid. The capabilities btrfs brings to the table are really useful for distributions, which is why there's always a desire to make it default, but the tools and performance may well need to catch up before this is plausible.
Originally Posted by mayankleoboy1
Btrfs snapshots alone would really simplify the murky waters that are Linux system restore utilities right now. A decent UI on that, scheduled snapshotting, and easy restore would kick the crap out of other options.
Originally Posted by AdamW
Tangentially, I just had a reaffirming interaction with btrfs. My main machine lost power, btrfs had a superblock go bad and would segfault on boot, from my recovery disk brfsck --repair fixed it easy. Once that tool becomes the mainline fsck.btrfs it would have recovered no problem. Promising!
leave it optional, please
I've been running btrfs on Fedora 17 and 18 on a couple of machines, I concur that it is clearly slower than ext4, as benchmarks show. Not to mention less stable.
Performance with virtual machines or anything that requires random writes inside a file is horrible. In Fedora 18 I retried this and it was still awful, moved my VMs back to disk partitions. This is being actively worked on, but for now btrfs is only performant for operations where COW makes sense. Database performance is usable for development only -- not for anything that approaches real production throughput.
Saying "performance is only good with compression on" is a cop out -- most data is not very compressible, particularly those requiring high throughput like database or audio/video. This succeeds in making btrfs look great in benchmarks using fake files only.
I also concur with the goal of waiting for Anaconda to stabilize. I installed Fedora 18 beta on a new laptop and only succeeded after Anaconda had hard crashed and forced reboot about half a dozen times. This is with very few partition options actually exposed in the Anaconda UI at the moment.
Also up for discussion would be whether there is a "best practice" btrfs configuration that should be used. i.e. should root be on a subvolume? What about /var and /home on separate subvolumes by default? should the @volname Ubuntu convention be followed?
A lot of things need to be shaken out before btrfs is a sane default. I would love it to be, but it's not near ext4 or xfs in terms of real-world performance, versatility and stability at the moment.