Abstract
Nowadays, the data fault-tolerant technique is to put data and erasure code in the same disk group. When a user would like to use another data fault-tolerant technique, he must use a new group of disks. In the situation, the disk utilization is very low, and a user cannot manipulate more than two kinds of data fault-tolerant technique with finite disks. This thesis constructs a hybrid data fault-tolerant system by distinguish the different location of files in a stand-alone computer. A user can decide how to store data elastically, and define the importance of data by himself, for getting a balance between the disk utilization and the execution speed of erasure code calculation.