16

The last real question I've seen on this topic is from about two years ago (is ext4 ready for production usage).

In the intervening time, how has ext4 improved?

XFS, JFS, and ext3 are the standby reliable choices. I've only used ext4 on recent Ubuntu test/dev environments, and haven't seen any issues - but they're also low-use workstations, VMs, and throw-away training environments.

How does ext4 stack up now that it's had [some] time to mature vs XFS and JFS (especially) from a speed and reliability standpoint?

Is ZFS a viable option (seeing as it's a fuse module, probably not for Linux - yet)?

warren
  • 17,829
  • 23
  • 82
  • 134

2 Answers2

6

ZFS on linux is unfortunately still not a viable solution, even if you dismiss the issue of being a FUSE module (which can seriously cramp performance on certain workloads). It simply isn't complete enough. Also, I don't think there's a debugfs for ZFS on linux, which is a serious negative.

debugfs is the traditional name for low level filesystem repair tool on unices. e2fsprogs include one for Ext2/3/4, XFS tools have xfs_db and others. Other filesystems, especially longer-existing ones like FFS and JFS have such tools too. It's basically a tool that allows you to read and manipulate the data on volume at much lower level, useful especially in recovery.

As for ext4, I'd suspect it's fairly usable in production, but I'd recommend actually simulating your workload on it. Be wary of various unsafe code paths in various applications that can corrupt the data depending on settings of ext4 (mind you, AFAIK those issues can happen in XFS and JFS as well).

XFS is still a good, stable solution, though I'll admit I moved from XFS to ext4 due to XFS' lackluster create/unlink performance. Still a very good choice if you don't have many small files being constantly created and deleted. Hard numbers can be taken from most benchmarks on the net. The slowdown is related to particular optimizations of XFS that cause certain journal operations to be quite slow (create/unlink). It's very fast in metadata access and read/write, though. Good choice for big files, IMHO (multimedia editing?).

Haven't really tested JFS, though I heard rather good opinions about it - just check first if it has a debugfs tool that you feel you can use reliably.

p_l
  • 216
  • 1
  • 5
  • 1
    Do you have any suggested links where we could read up more on the points you mention, e.g. *debugfs* and ZFS, XFS create/unlink performance? – ewall Jun 30 '11 at 12:38
  • so - do you have any reports or reliable comparisons to look at? (not that you're not reliable, but it'd be a bit nicer to take a rhat or canonical or gartner report back to management) – warren Jul 05 '11 at 20:35
  • 1
    @warren - I'm sorry for not responding earlier, but I wasn't monitoring the site recently. As for reliable comparisons, I can only point to various fs benchmarks on the net, including Phoronix. I can only assure you that their results fit *observed* behaviour of my system under both XFS and Ext4. As I mentioned in my answer, I can't reliably answer performance on the rest. – p_l Jul 27 '11 at 20:46
  • Has anything changed in the intervening period? I have a 10TB fs. At present it is xfs on LVM, but I'm looking at zfs. Will switch only if it's ready for production. – bsd Feb 01 '13 at 10:00
  • Since then, ZFSonLinux became a pretty good solution, and in fact I'm running it in production on my day job workstation. It works pretty good, but you need to be careful when setting it up. – p_l Feb 06 '13 at 07:57
2

ZFS on linux is now possible at a native level:

http://zfsonlinux.org/

ZFS includes a number of systems to ensure file system integrity, the most critical of them being zpool scrub, which does a checksum check and rebuild (if needed) of every single file.

For a production system today, I would not go with ZFS on linux. But if I had to store 40+TB of data in a few years, ZFS is where I would go.

n8whnp
  • 1,316
  • 7
  • 9
  • 1
    Don't say it's now supported. It's pre-alpha not ready even for SOHO: https://github.com/zfsonlinux/zfs/issues – poige Jul 28 '11 at 01:05
  • @poige - pretty sure he means that it is no longer *just* a `fuse` module – warren Jul 28 '11 at 11:55
  • @poige, thanks for the comment. What I was trying to get across is that it is possible to mount and use zfs natively. I indicated I would not use ZFS for linux on production systems yet, but if I had to build another large file storage solution, opensolaris + zfs is probably how I would do it. – n8whnp Jul 29 '11 at 16:08