smino Posted April 24, 2009 Share Posted April 24, 2009 Has anyone tested the BTRFS file system? I hear it is much faster and also performs software raid 0, 1, 10 right out of the box. It is included in the existing kernal builds. I would love to see a comparison between it and reiserfs. Maybe if we used BTRFS for the Cache drive? If anyone tests this with either seagate 500GB 7200.12 or 1.5TB 7200.11 Drives let me know. Those are currently the fastest SATA2 7200RPM drives, just as fast as the VeloRaptor 10K drives. http://www.linux-mag.com/id/7308/3/#comment-2488 Quote Link to comment
keland44 Posted April 24, 2009 Share Posted April 24, 2009 sounds like it would be nice to implemented it would be nice to see if Tom gives us the option to do this on the cache drive in the next beta how do you think this would be benefical if we used it on the parity drive? Quote Link to comment
perfessor101 Posted April 24, 2009 Share Posted April 24, 2009 Googling BTRFS turns up quite a few more hits than it did last year when I first heard about it ... If you've used Gentoo before ... just unmask and compile a kernal (sp?) ... (http://forums.gentoo.org/) They have a few threads and there were quite a few people who tried it out ... I haven't really looked into it much lately ... it looks like it will be a really good file system when it comes to maturity ... It even includes a sector by sector checksum when storing files, so it can detect corruption earlier (strong focus on data security) On Wikipedia - http://en.wikipedia.org/wiki/Btrfs When it's ready I would vote for it Quote Link to comment
smino Posted April 24, 2009 Author Share Posted April 24, 2009 Since BTRFS does raid 0,1,10 you get the option of protection or speed, or both raid 10 (but requires 4 drives). BTRFS is built into the Kernal. The question is: Is it compatible with unRaid. Unfortunately, I do no think it will be, something about the md if I recall. But if one of the developers wants to test it, and it turns out to be a performance advantage, then who knows. I was just curious to see how many people knew about btrfs. I know it is still fairly new/experimental, but stable enough to be included in the Linux Kernal. It competes with ZFS. As far as the parity drive goes, I think the parity rebuild will depend on all the drives and controllers driving them. The slowest one "say a PCI promise card 4 way" will slow down your parity rebuild. So it would not help in this case. However, you could have two 1TB drives raid 0 for parity, instead of having to buy a 2TB drive for parity and a 2TB drive to be used for storage. I am sure the linux/unRaid guru's will put in there thoughts soon... Quote Link to comment
Joe L. Posted April 24, 2009 Share Posted April 24, 2009 sounds like it would be nice to implemented it would be nice to see if Tom gives us the option to do this on the cache drive in the next beta how do you think this would be benefical if we used it on the parity drive? The parity drive does not have a file-system. Never did, never will. It is simply a collection of bits based on the data on the other drives. Don't hold your breath waiting for BTRFS, since Tom's foremost policy is "do not risk the client's data" and BTRFS is not anywhere near close. Btrfs is under heavy development, and is not suitable for any uses other than benchmarking and review. The Btrfs disk format is not yet finalized, but it will only be changed if a critical bug is found and no workarounds are possible. From what I see there, they are up to version 0.18 (as of Jan 2009). I'll wait until they get a LOT further on before trusting any of my data to it. Its focus seems to be how a "point-in-time-snapshot" of the data can be taken (for backup purposes) while the file-system continues to be used. This is a great feature for a online transactional database server, but not nearly as useful for a media server. (seeing how this is being developed by Oracle, you can see where they have their focus) It does this by making a "copy-on-write" (translation, any write to the drive replaces the old data block with a new block, located ELSEWHERE on the disk. With a lot of activity, this will result in a lot of fragmentation...) Joe L. Quote Link to comment
keland44 Posted April 24, 2009 Share Posted April 24, 2009 oh okay that's coo thanks for the clarification Joe but your right it will probably be a while before this does if it does get implemented. It's nice to wonder about though. Quote Link to comment
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.