I've had bad experiences with data loss on XFS. Personally, I like ext4 for my system because it's fast and reliable. I played around with JFS for a while, testing it, and I found that, while slower then some alternatives, it was ridiculously reliable and performed fsck's to recover data in mere seconds. I could not really get that FS to fail.

On my systems, I'll use ext4, but if i ever start a home server, I'll definitely use JFS.

Oh. Ignore me then.

(Do you like JFS by the way?)

i stick with ext4 for the moment.
It shows up that it's pretty fast in benchmarks , it's journaled filesystem and also promises fast fscks
Since it was considered stable since 2008 and now it's 2010 well ... at least sticking with it will help me have the bugs solved faster (mainly because many use it)
+1 ext4

Oh, I never heard of a filesystem that only works in a 64-bit OS, and I always have some Debian flavors in a VM, so I just tried it. I actually have never really used it.

lots of ppl are voting for ext4 here, that is seriously making me reconsider using it. but what data loss did you experience with xfs leopard? and how did it happen?

This is why I try not to post in threads like this, because all people care about is what the majority uses. No logic, no reason, just bandwagon.

Yes -- I've posted before about JFS' lack of popularity being caused by its, er, lack of popularity.

choosing file system is depend on the OS and the way we use the Hard disk (HD) to store data.

You can use ext3/4. But if you are going to use Linux then sure you can use ext3/4.

Also remember one thing the speed of the file system is dependent on the concept called journals. Today in field almost all the popular file system has the concept journals.

reg

thanx H_TeXMeX_H you're quite critical the thing i like, but you've been a little bit harsh on me! i just said that it makes me reconsider using not actually use right away. but what is eye catching is your opposition, you're almost crusading against ext4 which makes me ask why?

actually forums to me are more of an active poll (may be to others too) so this is why i also care about numbers, you know.

I don't understand it that way; it sounds more like crusading against decisions made on the basis of popularity rather than weighing the technical pros and cons.

In the world of computing, sometimes popularity is important -- a popular software is more likely to:

• Have been thoroughly tested and debugged.
• Continue in existence.
• Have good technical support resources.
• Have a good tool set (example: no JFS support in Knoppix).

That said, JFS and XFS are not sufficiently unpopular for those considerations to weigh heavily.

Like others have said, popularity is an important consideration if you look at why a piece of software is popular. ext3 and ext4 are both reliable and efficient filesystems IMO. I would compare the two and see which performs better and fails less (not that I've ever had a failure with either of these filesystems).

i totally agree with you catkin, and you cannot be more correct about that subject. also H_TeXMeX_H posted a couple of links that i found very interesting yet outdated, so i managed to do some research and found this http://www.nobius.org/~dbg/practical...tem-design.pdf which seems informative from the TOC. also i found http://homes.cerias.purdue.edu/~flor...r/reiserfs.php, and some other papers and still searching. but really what i would like to know about a fs is where it fails rather where it advances, this is how i can make an informed decision.

also the big new hdd is going to empty the smaller ones, and give me the chance of goofing around with different filesystems/operatingsystems.

Thanks for sharing the Practical File System Design PDF, a lucid and interesting read. I was confused by "In practice, the speed of CPUs so far outstrips the speed of disks that NTFS sees little performance difference in accessing compressed or uncompressed files". Isn't the opposite true in theory, if not in testing? Compression reduces the amount of (slow) disk I/O and imposes relatively little load on the (fast) CPUs -- so reading compressed files ought (TM) to be significantly faster ... ?

oh common don't hit the gas like that, i just was showing what i found. i never assumed that no one read it yet!