Beyond the Ratio: Smarter Compression for Object Data
Dell, Wednesday, October 28th, 2026
Dell explains how ObjectScale lets teams tune compression per bucket and measure the savings as AI data volumes grow.
Anahad Dhillon, who owns Dell's object storage roadmap, argues that data-reduction ratios only mean something in context, because backup deduplication and object compression work on very different data.
ObjectScale offers up to 9:1 compression of object data in Dell testing, with five modes (ZSTD, LZ4, Deflate, Snappy and none) set globally or per bucket, and a threshold skips compression when expected savings are low, as with already-compressed media.
He suggests starting policies for logs, training data and ingest, and points to dashboards that show savings by bucket.
Compression is presented as one part of a wider efficiency plan alongside erasure coding, space reclamation and media choice.