Hermitian codes in distributed storage systems with optimal error-correcting capacity

Maximum distance separable (MDS) erasure codes are widely used in distributed storage systems (DSS) for better storage efficiency and protection against Byzantine attacks. In this paper, we aim at enhancing the error-correction capacity of DSS in a hostile network. Firstly, we apply Hermitian code in DSS and presented a special placing mode for the encoded symbols. A reconstruction algorithm in error-free network is given. Next we show that the burst-error-correcting algorithm by Ren can correct more errors than Reed-Solomon code. We proposed an erasure rollback strategy in decoding. The new reconstructing algorithm improves both the lower and upper bound of error-correcting capacity. It has better computing complexity than Reed-Solomon code with the same storage efficiency.