Abstract

Btrfs is a relatively new filesystem, developed with the intention of becoming the default choice for Linux systems.

Whilst not yet finished, it introduces some exciting features not found in other filesystems, such as snapshots, storage pooling and data checksumming. In this talk, we shall look at some of these features, and the details of how they are implemented. Of particular interest is the Copy-on-Write system, and the novel use cases this enables, such as live snapshots of a running system. These features will be illustrated by live demonstration.

We will explore the technical details of how Btrfs is implemented, which differ significantly from other common file systems such as NTFS and Ext3. We will also consider the issues caused by the design of Btrfs, such as difficulty in determining free space.

Finally, we shall consider whether Btrfs is ready for production use.