The next generation of the extended file system; ext4, provides improved scalability, reliability, and considerable new functionality. This article covers ext4 Functionality, scalability, and performance. It's an easy read to get to know the latest and greatest Linux file system.
Post a Comment
Member since:
2006-02-15
More or less all the things mentioned in the article are things that benefit systems with high amounts of important data and very large storage systems; higher precision with time stamps, support for ever larger files and filesystems, the new syscall for allocating a large sequence of contiguous blocks for specialized applications....
But from the point of a regular user the only benefits ext4 bring is slightly more error-resistance and online defragmentation, though it seems you actually have to manually start the defragmentation with another utility. Neat feature, but it also remains to be seen if any distro is actually smart enough to add an easy GUI for doing that. Anyways, ext4 just doesn't seem to mean much for home users. Being able to create backups on filesystem-level would have been very useful if coupled with a good GUI, but I guess we'll wait for Btrfs to bring us that feature.
Member since:2007-09-08
On x86, ext3 can only handle volumes up to 8TB, and 2TB drives are already available. Give it about three years, and hard drives available in desktop systems will be banging against the upper limit for an ext3 volume.
I already have a 1TB hard drive and, while ext3 works fine on it, having to fsck that drive would be a nightmare. It'd easily take a couple of days.
Obviously ext4 does absolutely nothing to address my main concern with that amount of data (which is data integrity), but I guess we'll have to wait for btrfs for that one as well.
Member since:2007-03-13
I believe ext4 brings speed improvements over ext3 as well. Don't have the link handy giving the benchmarks. Pretty sure it was posted as a story here though.
I think this was it:
http://www.phoronix.com/scan.php?page=article&item=ext4_benchmarks&...
Edited 2009-02-23 15:45 UTC
Member since:2005-07-06
RE: Improvements for the average home user?
Member since:2007-08-11
While one should applaud how far ext has come since the FFS days, it seems hard to get excited about new features like improved timestamp resolution and checksummed journals when other free filesystems (e.g., ZFS) have existed for years that either already have these features or obviate the need for them, in addition to having modern features like snapshots and built-in compression.
Member since:2006-12-04
Member since:2006-02-15
Whoooosh.....right over your head. You are now talking about file systems. The OP'er was for some reason talking about the desktop. I agree, mostly, with you. But, I still say the OP'er is way way off base.
I beg your pardon, how exactly am I "way off base"? Yes, I know Linux is used on servers and ext4 does bring them some good features, but I was just wondering if it does bring anything for those who use Linux on the desktop. There actually are nowadays a load of those, too.
I was also pointing out that while the online defragmentation feature is a good thing to have, it needs a GUI tool in order for any desktop user to actually use it. Would also be interesting to know how much it affects performance and if there's any way to give it higher or smaller priority. That'd be useful on both server and desktop uses.
RE[5]: Improvements for the average home user?
Member since:2005-07-24
I'm pleased with what ext4 is. It was never expected to be btrfs. It was expected to be a last evolution of the mature and rock solid stable ext series of Linux filesystems. The new features add to its solidity, make it even more fragmentation resistant, and ward off some impending "barriers". Like one commenter has mentioned, 8TB and 16TB limits are about to get embarrassing.
If nothing else, the fast fsck will be a noticeable benefit to desktop users. I just did an fsck on a 500GB ext4 filesystem which was over half full, and which contained about 3 million files. 2 minutes and 47 seconds. I didn't do a side by side comparison. But I know from experience that ext3 would have taken a *lot* longer. A nice improvement considering that most of the "choice is good" Linux distro maintainers don't see fit to give desktop users a choice to break out of the automatic forced fsck's that occur by both mount count and time interval. I know I've sat through way too many, having forgotten to tune2fs away the silly mount count trigger.
Edited 2009-02-25 05:01 UTC
