First tests with btrfs on ΔV3

Btrfs is new file system being developed. GPL licensed, this is what the page notes:

Btrfs is a new copy on write filesystem for Linux aimed at implementing advanced features while focusing on fault tolerance, repair and easy administration. Initially developed by Oracle, Btrfs is licensed under the GPL and open for contribution from anyone.
Linux has a wealth of filesystems to choose from, but we are facing a number of challenges with scaling to the large storage subsystems that are becoming common in today’s data centers. Filesystems need to scale in their ability to address and manage large storage, and also in their ability to detect, repair and tolerate errors in the data stored on disk.

5 thoughts on “First tests with btrfs on ΔV3”

  1. Thanks for these numbers! I’ve been dying to try btrfs but I wanted to wait for it to stabilize a bit. From your numbers it looks pretty good. Can you post a quick summary/comparison to other file systems? I’m curious if a pure RAID-0 with btrfs is faster than xfs.
    Any thoughts about how btrfs compares to zfs in terms of features? I know you’re not supposed to post performance of zfs on OpenSolaris, but if you could “hint” at a comparison 🙂 I’m just curious if btrfs is a true competitor to zfs in terms of features and performance
    I haven’t really played with ceph. I’ve read a little about it though.
    A combination of btrfs and GlusterFS might be really interesting. Any thoughts about testing this combination?

  2. @Jeff
    Haven’t done a side by side between zfs and btrfs. Will do eventually. In either case, it is moot, as most of the world can’t use zfs without using Solaris (which despite protestations to the contrary from various sources, isn’t likely to ever happen). btrfs is GPLed and in the kernel now. Ceph is GPLed, though it might not get into the kernel for a while (if ever). GlusterFS is quite good.
    (stepping back) I think the issues that arise are, as file systems get “huge-er and huge-er” 🙂 exactly what should file systems be like going forward? Some ideas from recent past history make no sense for very large data and permanent storage. Some make a great deal of sense.
    GlusterFS looks like it addresses some of the needed issues in a “global” sense, and does so, from my perspective, better than a number of alternatives. Btrfs addresses some of the local storage needs quite effectively, and from a brief feature list comparison, looks like (in conjunction with the data integrity patches in the Linux kernel), will be able to address a wider set of potential problems with data corruption than some competitors in the market now. This is a guess, but we will see what happens.

  3. Hi Joe, looks like those numbers have dropped off the blog ?
    I’ve been playing with btrfs for a year or so now and it’s really quite fun, though the fact that different chunks in the filesystem can have different RAID levels for instance takes a little bit to wrap your head around (and can lead to confusion looking at the output of mkfs.btrfs and df).
    I’ve been running it in anger on my laptop with an SSD drive for the past few weeks (after some successful tests) using its internal RAID-1 implementation over two partitions in case of disk errors and it’s not missed a beat.

  4. Forgot to mention that there is a project called CRFS, also from Oracle, which is intended to produce a remote filesystem based on btrfs’s internals. It’s pretty quiet though!

Comments are closed.