The Btrfs Community

Filesystems span many different use cases

Contributors include Oracle, Redhat, STRATO, SuSE, Fujitsu, Intel and many individuals

Broad community ensures Btrfs is full of unique features

Copy on write metadata and data
Efficient writable snapshots
Internal RAID with restriping
Online device management
Online scrub
Transparent compression
CRCs for data and metadata

Storage Management

Btrfs dynamically allocates chunks from each device

Metadata

Data

Chunks from different drives are combined to create RAID

Online resize and restriping operate by relocating chunks

Benchmarks

Random 4K Writes on RAID0

Btrfs

Ext4

Btrfs allocations and writeback are optimized for SSD storage

Support for SSDs as a front end cache is coming

SSDs and other advanced devices are notified via TRIM commands when blocks are no longer used

Trimmed blocks are ignored while cleaning segments for reuse

Scrubbing verifies data and metadata integrity

Duplicate copies are checked in parallel

Seed devices start with a readonly image

Writable devices are added to create new filesystems

Worker threads spread CRC and compression across CPUs

Recent changes address context switch latencies

New in 3.5

Device IO error statistics

Much lower IO completion latencies

Tree lock contention fixes

Improved backref walking

Planned for 3.6

RAID5 RAID6

More flexible RAID1, RAID10

SSD caching layer

Subvolume quotas