Guest post by The latest Fedora development snapshot makes Ext4 the default file system and adds experimental support for the next generation btrfs filesystem. “According to current plans, version 11 of Fedora, which is expected to arrive in late May, will use Ext4 as its standard file system. That’s what the Fedora Engineering Steering Committee (FESCo) recently decided, following a heated discussion in an IRC meeting. If however Ext3’s successor encounters big problems with the pre-release versions of Fedora 11, the developers will dump that plan and revert to Ext3.”
Although it’ll undoubtedly increase the testing and improve the adoption rate and the driver itself, I can’t not to notice it’s typical for FOSS projects to quickly embrace fresh and yet unproved technologies. That’s just how it works here, but it might leave a bad taste for some unsuspecting users.
I wish the best luck for ext4 project and a painless migration for all of us
not that much of a rush. ext4 has been an option since fedora 9 (maybe 8? but i dont think so), so its not that new. i am interested to see how it runs when 11 gets to a stable beta.
not a troll a genuine question:
why should I care what file system my computer uses? what difference does EXT4/EXT3/Fat32/ntfs really make.
As an end-user will it really matter?
As a JOE you shouldn’t care. You might experience slightly better performance with ext4 and hopefully no data loss. In general the distro developers should choose the “best” defaults for you. Which is why I’m a bit worried, but curious at the same time.
Yes it matters.
I have been using Reiser4 on my laptop for years now. It’s performance has been much better than Reiser3 or EXT3.
Improved disk layout via delayed allocations and the like mean that a seek-challenged drive like a laptop’s can write new files to disk much more quickly.
Improved directory indexing means that email files in very large maildir stores (like what Evolution uses for IMAP cache) can be accessed very quickly. This also improves speed for GIT operations.
In my case, Reiser4’s full data journaling has helped a lot when -mm series kernels go bad and I have to force power off to get the laptop back. I never ended with screwed up email directories with Reiser4.
It has a significant effect on performance – how the filesystem works can affect how well it works on operations involving large amounts of files, either sequential or randomly ordered. It also has implications for data safety: properly implemented journalized filesystems are far less likely to lose data in a hard power-off than others.
One interesting upcoming area is performance on SSDs. Using ext4 and, to a greater extent, btrfs will lead to significantly better performance on SSDs than ext3.
In addition to what the others said already, I’ll give you a simple example. Do this test. Pick a file that is larger than 4 GB. Copy the file in a fat32 partition. Check the size of the file after copying it in the fat32 partition. Is it the same?
Except you can’t copy a >4gb file on fat32 partition. But yes, large files do take more space on fat32 and ntfs partitions than on ext3. Not by much, though.
It does.
E.g. A brts aware installer can revert to a previous version of your root file-system, in-case something goes horribly wrong during an upgrade / installation.
No more “I just installed an upgrade and it bricked my OS”.
– Gilboa
Yeah, but ext4 was unreleased back then. Using ‘community’ distros is like a constant beta testing. Which is not necessarily bad, when you know what to expect.
Ext4 wasn’t “unreleased”. It was marked as a development filesystem and has been merged in the Linux kernel with that tag a while back. In Fedora 9, you have to pass a explicit boot option and was clearly noted as experimental so only those who understand the risks will be involved in early testing.
Red Hat has a few full time kernel developers working on Ext4 and btrfs so I am sure they understand what is involved and can get a headstart.
‘Fresh and unproven technologies’, mate, it is a natural evolution of EXTFS, it is hardly an ‘unproven technology’. Its no more ‘unproven’ than when Microsoft updates NTFS to offer more features/enable features that weren’t available in the previous Windows generation.
EXT4FS is hardly what one could deem a ‘revolutionary’ leap forward, it is a conservative movement forward that isn’t disruptive by any stretch of the imagination – and the changes made aren’t exactly so risky that it makes it ‘unproven’.
Using the same benchmark then maybe we should call all of Microsoft’s file system drivers so far as ‘fresh and unproven’ given they suddenly appear or features suddenly added to NTFS without a ‘adequate testing period’ (what ever that means).
I wouldn’t say unsuspecting, at least not in all cases. Take me for example, I use Fedora explicitly to dance as close to the leading/bleeding edge of things as I can reasonably get without resorting to compiling things by hand. If I wanted something more conservative I’d probably be using a RHEL variant like CentOS or maybe even Debian.
This is great news!!!
Will mean that there is a high chance that RHEL6 will use ext4 as a default file system.
If you are in a corporate environment then you will use RHEL6 rather than Fedora because as mentioned in a few comments you have to treat Fedora as a beta type of operating system.
RHEL 6 will have Ext4. This was confirmed in the Fedora Eng team meeting by Ric Wheeler, kernel developer and filesystem team manager for Red Hat
http://jwboyer.fedorapeople.org/fesco/fesco-meeting-2009-01-16.txt….
I bet in future btrfs will rock the world. Right now it is incomplete and buggish but ALREADY outperforms virtually any competitors on most patterns. It will knock down competitors when ready, IMHO.