Messages in this thread | | | Date | Thu, 30 Mar 2000 14:00:32 +0400 | From | Hans Reiser <> | Subject | Re: Is ReiserFS really a journaling file system, or is it really just a synchronous-metadata file system like BSD FFS? |
| |
ReiserFS is a journaling file system. There are two types of journaling file systems: data journaling, and metadata journaling.
ext3 is a data journaling file system, but Stephen intends to make it a metadata journaling file system.
It is less work to author a data journaling file system than a metadata journaling file system, since you don't have to agonize over what needs to be journaled. The problem is that data journaling means copying twice as much data for large files, and most users say they want data journaling right up to the point at which you tell them it will halve their performance for large files writes. At that point most users say they prefer metadata journaling. This is why we did the extra work of making ReiserFS metadata journaling, and why it is so much faster than ext3 (and will be until Stephen has time to make it metadata journaling). In a few months we will do a new form of journaling we are inventing called wandering logs, and then you may finally have decent performance with data journaling. We will make data journaling a mount option, after Chris gets his very overdue vacation which will be after 2.4 ReiserFS is sent to Linus.:-/
Hans
ncm@nospam.cantrip.org wrote: > > I have seen the ReiserFS described as a "journaling filesystem", but > in examining the project web page, it appears not to be. In fact, I > see described there a "journaling patch" which seems to journal FS > metadata, but not file contents. In particular: > > This is metadata logging only. If you hit the reset button in the > middle of a kernel compile, you will end up with a bunch of > invalid .o files, but the directory tree will remain valid. If > you run sync a few times and then hit the reset button, data and > metadata will both be valid. > > http://devlinux.com/projects/reiserfs/jrnl/ > > As I understand it, a journaling file system is one in which, if you > pull the plug at any point, all the metadata _and_ file data accurately > reflect a recoverable self-consistent state at a some instant. IIUC, > it is what is promised by Veritas and the IBM and SGI systems which are > being groomed for inclusion. Not to journal file contents somewhere > seems to make such semantics impossible. > > Is it that the file system layout itself supports journaling, but > that the VFS layer doesn't support the semantics needed really to > get it? > > (This also begs the question, if terminology is so abused here, is ext3 > really a journaling fs? (And, is ext2 a "journaling fs" just because > the ext3 patch exists?)) > > I hope my impression is wrong, and that in fact all these file systems > provide similar recovery semantics. Can someone answer authoritatively? > > Nathan Myers > ncm@nospam.cantrip.org > > p.s. I know ReiserFS has all kinds of cool semantics in addition to > traditional file-system stuff; I'm just asking about its recovery > characteristics. > > p.p.s. Sorry about the title, it was my impression that all > ReiserFS-related postings have to have long titles. :-) > > - > To unsubscribe from this list: send the line "unsubscribe linux-kernel" in > the body of a message to majordomo@vger.rutgers.edu > Please read the FAQ at http://www.tux.org/lkml/
-- You can get ReiserFS at http://devlinux.org/namesys, and customizations and industrial grade support at reiser@idiom.com.
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |