Work out block size overhead instantly with clear inputs, formula shown and shareable results.
Files occupy whole blocks, so a 3,400-byte file in a 4 KB block filesystem wastes 696 bytes of slack. Multiplied across hundreds of thousands of small files that becomes significant, and metadata adds more. Larger blocks improve sequential throughput but worsen slack, which is why 4 KB remains the general-purpose default.
Block allocation
blocks per file = ceil(size / block size); allocated = blocks x block size + metadata; slack = allocated - actual size
Btrfs and ext4 with inline data can store very small files inside the inode itself, and ZFS uses variable block sizes, all of which reduce slack considerably.
For large sequential files such as video or database extents, where fewer, larger allocations reduce metadata and improve throughput.