Locality‐Seeker: Exploring Potential Locality for Efficient Data Reduction in Backup Storage

ABSTRACT Data deduplication and delta compression are widely used in backup systems to improve storage efficiency by eliminating redundant data. However, while effective, these techniques introduce data fragmentation, which can significantly degrade restore performance. MeGA, a state‐of‐the‐art framework, addresses this fragmentation issue but fails to exploit redundancies across non‐consecutive backup versions due to its limited focus on adjacent backups. We present Locality‐Seeker, an enhanced backup system that extends MeGA by indexing adjacent chunks from earlier backups, enabling it to uncover and eliminate cross‐version redundancies. Our evaluation on real‐world datasets shows that Locality‐Seeker improves the compression ratio by up to 50.86% compared to MeGA, while maintaining comparable restore performance. These gains come at the cost of only modest increases in memory usage and maintenance overhead.