Building a High-performance Fine-grained Deduplication Framework...
summary
MeGA is a deduplication framework that applies selective data encoding to skip encoding chunks in containers rarely serving as bases, reorganizes storage into a delta-friendly data layout so restore-required chunks sit compactly, and uses Always-Forward-Reference traversal plus delta-aware writes to avoid repeated container reads. Implemented and evaluated on four backup datasets (VM images, website snapshots, source trees, synthetic), it achieves 4.47–34.45× faster backup and 30–105× faster restore while keeping a very high deduplication ratio.
AI-written summary · about