What is Digg?42 Comments
- 2gooder, on 10/12/2007, -17/+45You know I've always been a ReiserFS fan myself... that was until I realized that what ever file system I'm using on my Linux Desktop has absolutely no bearing on my computing experience and the work I get done. Now I live a large portion of my life outdoors, and with girls. I haven't looked back since. Come join me.
- nacs, on 10/12/2007, -1/+25"Who the hell more than 8TB in a single partition? Split it up, damn!"
This sounds a bit like the mythical Bill Gates quote "640K ought to be enough for anybody". My first PC had 33MB of disk space. My current system has 1.2TB.
Having more than 8TB+ per partition will eventually become commonplace. It's just a matter of time. - drag, on 10/12/2007, -0/+18Ya.. It's too bad about reiserfs 4.
Reiser is realy caustic sometimes and he did drop the ball on Reiserfsv3. He was trying to tell the kernel developers that it was depreciated and all that.. which of course with a file system that is just absurd.
But reiserfsv4 realy has neat features.. Especially for things like it's plugin systems. For instance encryption and compression plugins would be great for mobile media. No other Linux (usable) file system offers live read/write file system compression.
ZFS is cool, but you have to keep in mind that it's license is currently incompatable with Linux (pretty much out of design by Sun) and if Sun was to release it it may not get that much attention from developers.
The Linux kernel people like to solve problems universally.. So we have all these layers of software that you put together in different ways to solve problems. You have the MD software raid stuff (which is realy realy fast for software raid, often times out performing real hardware raid solutions), the DM stuff. LVMv2 and it's clustering counter part CLVMv2. You have lustre distributed file system which is closely related to ext3 developments. You have OCFSv2 from oracle for clustering. GFS and assorted stuff here and there.
Linux file system and such is very varied and capable. Having ZFS for Linux would lead to a LOT of redundant functionality. More likely the kernel developers would rather take portions of ZFS and abstract them out so that they benifit all file systems by being used in the Linux-VFS stuff.
But it's hard to say, I am not a kernel developer.
Then we have stuff like stackable file systems.. (UnionFS for instance) and user space file systems based on Fuse, for instance one I like a lot is sshfs, which is file system based on ssh. Very usefull and fast.. beats out NFS on slower links, supports seeking through files and such... Also you don't need to be root to setup file systems and mount them. Another is Encfs, which allows users to create encrypted file systems stored as directories.
But seriously.. I would love to see Reiserfsv4 in the kernel. It's a sad state of affairs that it's not in. (Although I would probably rather trust my data to Ext3.) - inactive, on 10/12/2007, -10/+27I know, I hate being unable to use my EIGHT TERABYTE RAID SYSTEM ON EXT3.
Who the hell more than 8TB in a single partition? Split it up, damn! - carpespasm, on 10/12/2007, -2/+19not until microsoft pulls their head out of their ass and play nice with the other computers of the world
- dougmc, on 10/12/2007, -2/+17`that was until I realized that what ever file system I'm using on my Linux Desktop has absolutely no bearing on my computing experience and the work I get done'
Nice troll. Yes, for most uses, your precise filesystem (as long as it's ext3, xfs, reiserfs, jfs, ufs or something similar) doesn't matter, but when you start putting millions of files into a filesystem that's hundreds of gigs in size, the filesystem chosen starts to matter.
But you enjoy your time outside with your girls. I'll enjoy my time outside as well, and my daughter and son like to come out with me too. But for work, if I'm setting up a heavily used server (it does pay the bills, after all) I've found that the filesystem I choose does make a difference.
(And I've found that ext3 tends to underperform xfs and reiserfs. ext3 satisfies the KISS principle, but beyond that, I'm going with xfs now.) - btipling, on 10/12/2007, -3/+16You obviously never had to sit through a startup where it checks the entire filesystem each time the computer shuts down unexpectedly with ext2.
- VorpalK, on 10/12/2007, -2/+14I know the reason I personally didn't get behind Reiserfs 4 is in no small part influenced by Hans Reiser's inflated ego and attitude. I don't want the man's file system to be widely used. The fact that the only Linux partition that I have EVER lost data on has been a ReiserFS 3 partition. I'm sure it's just a coincidence that old Hans only answers paid support questions.
ZFS looks very promising though. - raid517, on 10/12/2007, -5/+17Girls... yeah, we believe you....
- inactive, on 10/12/2007, -0/+11By that time, we'll all probably be completely happy using ZFS instead.
- MacSuxWindozSux, on 10/12/2007, -1/+11Rich people with hardware you don't even have the option to buy need this.
I can see the NSA saying... wow finally.. cant wait. - raid517, on 10/12/2007, -2/+11Thank god. Maybe the OSS community can get behind this in the way they didn't get behind reiserfs 4?
- danglerman, on 10/12/2007, -5/+13This is unexpected, i always thought ext2 and ext3 were great filesystems. But, then again, we will be getting bigger drives. Bigger Drives=Larger Filesystems
- RetroRufio, on 10/12/2007, -3/+10point well taken :)
- idonthack, on 10/12/2007, -0/+6@Eztli
Large organizations with large databases. - diggAddict, on 10/12/2007, -0/+6Right on brother - Solaris 10 for me!
- dougmc, on 10/12/2007, -1/+6How would this have any effect on NTFS?
Unless Windows adds ext4 support (which seems unlikely) I see how this will have any effect at all.
(Paragon's `Mount Everything' does allow you to mount ext2/ext3 under Windows, and while it's not perfect, it seems to work reasonably well. I doubt it can do the Windows boot drive, but it can do the data drives, which seems a reasonable compromize for a dual boot machine, far better than using fat32.) - jvimal, on 10/12/2007, -0/+4I hope compression would be a part of the ext4 file system...
- fanboydcs, on 10/12/2007, -2/+5Wrong, Speed enhancements alone make it better the Desktop, Servers need stablity thats why ext2 is still used on them.
- Obsidian743, on 10/12/2007, -2/+5Am I the only one amused by the Microsoft SQL Server add that plays just below the article?
- dougmc, on 10/12/2007, -1/+4Why? Compression at the filesystem level isn't actually that useful.
It tends to kill performance, and certain calls (like mmap()) often don't work on a compressed volume, which would break lots of applications. Also, at least for your typical home user, most of your data stored is already compressed and wouldn't compress more, and for your typical server user, the performance loss would be the killer.
Stacker had it's day. This day has passed. - i440, on 10/12/2007, -1/+3I hope that the filesystem can be resized.
I remember trying to resize a Debian parition on Ext3, and those memories were not good. - RetroRufio, on 10/12/2007, -5/+7Perhaps this will pave the way for a NON-pain in the ass NTFS filesystem compatability.
- doldr, on 10/12/2007, -0/+2I've resized many ext3 filesystem and it has so far never failed (see ext2online)
- Bogtha, on 10/12/2007, -0/+2Compression can actually help performance in some cases. Remember that CPUs are a hell of a lot faster than disks, and most of the time they are sitting idle. If you can compress something so that you don't need to write as much to disk, then unless the CPU is the bottleneck in your system, you've sped things up.
You seem to be thinking about an all-or-nothing approach to disk compression. It's perfectly possible to compress only certain files or directories, you don't have to compress the whole partition. /var/log, /var/mail, /usr/ports, /usr/portage and /etc would get good compression ratios. - ungamedplayer, on 10/12/2007, -0/+2Ok, assuming that people see big numbers like "128" and cool names like "ZFS", do people assume that it is a "good" filesystem. I know that Linux can not use the ZFS code, but I have not looked into the licencing of if it can be implemented or not.
Does anyone know ? Please back up your opinions with links. - dougmc, on 10/12/2007, -1/+3`but I don't think anyone with a desktop has the amount of storage that this file system is capable of.'
You never know. And perhaps nobody needs it today, but in a few years somebody might need it, and the ideal time to make sure your filesystem can scale in some way is _before_ it's needed, not after.
And also note that not all computers sit on the desktop. Many sit in the closet or data center and are called `servers', and often these computers have a lot more hardware than what you've got on the desktop. - bigtomrodney, on 10/12/2007, -0/+1ext3 was only a few patches to ext2 - journalling to be specific. The ext2 drivers can read ext3 safely, but just won't build a journal.
- doldr, on 10/12/2007, -0/+1What now seems to be turning into ext4 was originally planned to be just a few more patches on ext3. So in that sense it's not quite a new file system, it was just decided that ext3 should be left alone :-).
Also, some of the patches that makes up ext4 will be of use to people with "normally sized" file systems too. So it's not a case of 100% who-cares-since-I-dont-have-an-8T-fs. - inactive, on 10/12/2007, -0/+1Reiser4 isn't that great for desktop systems. It might have great speed and lots of features however it can use a lot of cpu in the process. I also had 2 filesystems die completely on me using reister4 and had to run the disk fixing stuff quite a few times. This was a while ago so maby its stabler now (It was just after amd64 was supposed to be usable although only one of my system was actually using amd64).
Since there are 16 PetaByte drives being invented having 8TeraBytes isn't that unlikly and considering how many systems still use ext2 (also consider ext3 is almost exactly the same as ext2) getting support now cound be smart. Although with such large drives there will probally be new filesystems designed specifily for drives that large, such as never bothering to delete and storing a record of all changes so you can revert to an older file or a brach of a file. - Snarfy, on 10/12/2007, -1/+2For those in the unix world it's calld Slowlaris 10.
- repruhsent, on 10/12/2007, -7/+7lol, what? I think you forgot to include something, like a point!
- ajs05, on 10/12/2007, -3/+3Who needs an 8TB FS? people doing HD-DVD and Blu-Ray encoding. Which right now
is not that many people but a year or two from now, may be you and I. - djironx, on 10/12/2007, -3/+2This is good for servers with really big storage capacities, but I don't think anyone with a desktop has the amount of storage that this file system is capable of.
- Urusai, on 10/12/2007, -5/+2Why bother extending such a naive implementation of the VFS? There are plenty of other file systems out there.
- tehmoth, on 10/12/2007, -6/+2guess ZFS is too hard to implement for them
- MrCoke, on 10/12/2007, -7/+3ext4? sweeeeet name, totataly un-expected. i thought it was gonna be Liger or something :P
- 0siris, on 10/12/2007, -13/+8Quoted for truth.
- fanboydcs, on 10/12/2007, -13/+4repruhsent, I just wanted to make digg users smile:)
- fanboydcs, on 10/12/2007, -15/+5I made dinner for the first time ever, I sit down to enjoy the Chicken and veggies I made and I see this tid bit of news. Man did that make my first ever dinner experience amazing!
But anyway, I have a ext3 system now and I was just thinking how much more robust it is than hfs + I truly think that the file system is the cause of harddrive failure. - repruhsent, on 10/12/2007, -17/+5lol
- repruhsent, on 10/12/2007, -24/+5How did this comment get dugg? It's a no-brainer that anyone could have thought of - it's common sense for *****'s sake.
Check out the new & improved 
© Digg Inc. 2009
— Content created and posted by Digg users is