Optimizing Storage and Computation Costs in Backup Systems: A Dynamic Decision Model Based on Information Entropy

Balancing processing time and storage space is a key challenge in backup system design, especially when data are heterogeneous. This study presents an entropy-adaptive backup strategy, called SmartEntropy, which uses information entropy to decide whether compression should be applied in order to reduce overall backup cost. In this study, three strategies are examined: full replication without compression, full compression applied to all data, and the proposed entropy-adaptive strategy. Experiments are conducted with different values of the time-relative cost weight W to evaluate storage usage and CPU time. The results show that the entropy-adaptive strategy generally achieves low storage usage with moderate CPU overhead. It outperforms traditional strategies when the cost-to-time weight is low to moderate, reducing the total backup cost by about 40–50% in the experimental setting. When the time-cost weight exceeds a threshold of approximately 0.5, the uncompressed strategy becomes more favourable because compression overhead dominates the total cost. Finally, the paper discusses the applicability and limitations of the proposed strategy and outlines possible directions for future improvement, providing practical guidance for backup policy planning.