Research Of The Reliability Of Network Storage System Based On Redundancy Mechanisms

Jim Gray, winner of the ACM Turing Award, introduced a new experience law: in the network environment, every 18 months the volume of data generated equivalent to the volume of data and ever. The explosive growth of information resources makes increasing demands for storage system in the storage capacity, data availability and reliability.Redundant mechanism is the key technology to improve the reliability of the network storage system, and therefore the focuses of the research are hardware-based redundancy as well as RAID-based and Erasurecode-based data redundant mechanisms.Hardware-based redundancy is redundant network interfaces that provide services for the system, making a number of links integration into of a virtual network channel. Through network interface and link redundancy, combined with detection of the failure and restore, to ensure the unimpeded access to the channel, it provides a reliable network I/O environment for the network storage subsystems.Raid-based redundant data distribution is through storage virtualization technology to spread over the network disk resources into a corresponding level of raid, redundant information and data through striping and interleaving-distribution, in case of storage node failure or loss of data can use redundant information to online restore lost data or critical failures storage nodes, which can greatly enhance storage system reliability. Given the current widespread Heterogeneous storage in network storage system, the data distribution of Heterogeneous disk environment is studied, and the optimal rate allocation with no capacity constraints and he optimal rate allocation with capacity constraints are introduced.(n,k) Erasure code is a highly efficient fault-tolerant coding, which encodes k-source data into n (n>k) of the data, using the arbitrary n-k data can reconstruct the original k data sources. By applying this highly efficient coding method to storage system, Erasure code-based data redundancy is introduced, and key technologies involved such as data read and written, failure detection and node redirection as well as the restoration algorithms are studied. The theoretical analysis concludes this redundancy scheme to enhance system reliability has more advantages: to obtain the same availability of data, Erasure code-based needs only lower redundancy degree; also in the same degree of redundancy, the scheme based on Erasure code has higher availability.