hesela.devBetrFS
BetrFS is a research filesystem that uses B-epsilon-tree indexes to organize file data and metadata.
The FAST 2015 prototype runs in the kernel and stores its index nodes in files on an underlying ext4 filesystem. Its filesystem interface and its index are distinct layers. [1]
The FAST 2016 paper calls the earlier design BetrFS 0.1 and adds late-binding journaling, namespace zoning and range deletion in BetrFS 0.2. These version-specific mechanisms address different costs; namespace zoning here is unrelated to drive zones or fabric access control. [2]
Relations
- related
- B-epsilon tree, Write-optimized dictionary, Buffered upsert, Full-path indexing, Range rename, Late-binding journal, Copy-on-abundant-write
Limitations
- Historical prototype evaluations do not establish present-day production readiness or a universal speed advantage. [1, 2]
Citations
- Jannen et al. BetrFS: A Right-Optimized Write-Optimized File System. FAST 2015, pp. 301-315, sections 4-6.
- Yuan et al. Optimizing Every Operation in a Write-Optimized File System. FAST 2016, pp. 1-14, sections 1-5.