二段重複排除方式による排除効率の改善

抄録

This paper proposes a two layers deduplication system for backup process in IT environment. The proposed system implements two chunk sizes in descendent way and activates them simultaneously in a sequence. They eliminate duplicated parts among backup target data in more efficiently than any conventional single layer deduplication system. The system weakens a substantial trade-off between the deduplication rate and the performance, which is defined as the number of chunks reduced. Comparing to a conventional approach the proposed system provides the equivalent deduplication rate with much less performance degradation or more performance with less compensation of deduplication rate. In addition, it achieves better stability of performance when small chunk sizes manipulate heavily duplicated data and densely changed data. The positive effectivenesses of the proposed system are evaluated quantitatively and are compared to a conventional approach by using a simulation. The results are certified with a prototype machine.