The BTRFS balance operation (btrfs filesystem balance) is primarily designed for data redistribution across devices in multi-device filesystems. While it rewrites data blocks through the allocator, it doesn't explicitly defragment files in the traditional sense. The operation processes entire chunks (typically 1GB) rather than individual files.

# Balance operation (whole filesystem)
btrfs filesystem balance start /mount/point

# Defragmentation (specific file or directory)
btrfs filesystem defragment /mount/point/path/to/file

The balance operation will relocate data within the same chunk if possible, rather than attempting to coalesce fragmented file extents. This means:

• Balance may incidentally reduce fragmentation when moving data between chunks
• Balance won't optimize file extent layout within a chunk
• Heavily fragmented files may remain fragmented after balance

Here's how to check fragmentation before and after operations:

# Check file fragmentation
filefrag -v /mount/point/path/to/file

# Compare before/after balance:
sudo btrfs filesystem balance start /mount/point
filefrag -v /mount/point/path/to/file

Use balance when:

• Adding/removing devices from multi-device filesystem
• Recovering from RAID degradation
• Reclaiming unused allocated space

Use defrag when:

• Specific files show high fragmentation
• Optimizing performance for frequently accessed files
• Preparing files for snapshots (reduces COW overhead)

Balance operations are I/O intensive and should be:

• Scheduled during low-usage periods
• Limited using filters (-dusage, -musage)
• Monitored for progress (btrfs balance status)

# Example of filtered balance
btrfs balance start -dusage=50 /mount/point

For optimal results on single-device systems:

# First balance to consolidate space
btrfs balance start -dusage=80 /mount/point

# Then defragment critical files
find /mount/point/path/to/files -type f -exec btrfs filesystem defragment {} +

Remember that defragmentation invalidates COW for modified files, so use judiciously on files that won't be snapshotted.

The btrfs filesystem balance command primarily serves to redistribute data across devices in a multi-device Btrfs filesystem. When examining its core functionality through a developer's lens:

# Basic balance command
sudo btrfs balance start /mnt/btrfs

# With filters (common developer usage)
sudo btrfs balance start -dusage=50 -musage=50 /mnt/btrfs

Contrary to some assumptions, balance doesn't inherently defragment files. The operation works at the chunk level rather than file extent level:

• Data is moved in allocation chunks (typically 1GB by default)
• File extents maintain their original fragmentation patterns
• The allocator may place fragments differently but doesn't consolidate them

Developers can verify fragmentation before/after balance:

# Check file fragmentation
sudo filefrag -v /mnt/btrfs/large_file

# Compare with defrag operation
sudo btrfs filesystem defragment -r /mnt/btrfs

For optimal filesystem performance, consider this workflow:

# First rebalance metadata
sudo btrfs balance start -musage=70 /mnt/btrfs

# Then defragment critical files
sudo btrfs filesystem defragment -r -f -t 32M /mnt/btrfs/database_files

# Finally balance data chunks
sudo btrfs balance start -dusage=80 /mnt/btrfs

Benchmarking shows different characteristics: