With the continuous increase of data, the data center that plays the role of backup is facing the problem of energy hunger. In practice, to reduce the bandwidth, the local data is synchronized to the data center based on incremental synchronization. In this process, the data center will generate a huge CPU load. To solve this pressure of the data center, first, we analyze the process of the Rsync algorithm, the most commonly used in incremental synchronization, and CDC algorithms, another way of chunking algorithm. Then we propose a data structure called Shadow Data, which greatly reduces the CPU load of the data center by sacrificing part of the hard disk space in the local node.