High availability mechanism for parallel file system

Parallel file system achieves a high I/O throughput by dividing a file into multiple blocks and storing them on multiple I/O nodes. However, the reliability and availability of the parallel file system is sacrificed for distributing file data over multi I/O nodes. A new mechanism named logic mirror ring was developed to improve the reliability and availability of the parallel file system. A logic mirror ring was built over all I/O nodes to indicate the mirror relationship among the nodes, i.e., each node maintained both its own data and the mirror data of other nodes. Moreover, the mirror depth can be adjusted to different levels based on the requirements of the reliability and availability. The effects of logic mirror ring on the reliability and availability of the parallel file system were evaluated by a Markov chain model. The mathematic solution shows that when the mirror depth is 2, applying the proposed mechanism to parallel file system, the average time of losing data is about 32 times of the original system.