A Research Of The Backup Technology Based On Erasure Code In Distributed Storage System

With the development of Internet and information technology, the number of data isgrowing exponentially. Facing growing data pressure, the traditional storage technologythat depended on the local server storage has been unable to satisfy the growing demandof data, so a lot of data centers based on distributed storage system are growing rapidly. Inorder to ensure the availability of data in distributed storage system, disaster tolerance isone of the key problems. It needs to use a variety of fault tolerant strategies to achieve thehigh reliability; especially the distributed storage system contains a large number of colddata. In view of the above questions, with the various erasure code schemes analyzed, theerasure code scheme of fault-tolerant strategy in the distributed system is put forward.After several erasure codes are analyzed, a backup scheme based on the distributedstorage system consist of a lot of cold data are introduced. Using erasure code as backuptechnology in the distributed system, could guarantee the reliability of the system, reducerecovery time and improve storage utilization and flexibility. Through the nodes dividedinto multiple virtual nodes, the load balance of system is better. According to the customerdemand, the backup servers select storage group which consist of virtual nodes. Aftersegmented and encoded, the user data could be stored in the virtual nodes dispersedly.Different user can choose different erasure code. When the data block broken, the systemcould repair it through decoding. The recovery time could be less by parallel decodingwhen nodes broken. In this condition, the distributed storage system based on erasure codecould ensure the reliability of system and reduce the recovery time when nodes broken.Based on design model, the erasure code scheme is simulated and tested by software.Through testing choose virtual nodes and data recovery, the results were counted. Theresults showed that this erasure scheme in distributed storage system could achieve highreliability and recovery efficiency. The storage utilization of system based on erasure codewas36%higher than that based on copies; in addition, this recovery time of this schemewas shorter. The CRS code and RS code were adopted mainly in test; the result showedthat CRS code was better than RS code in efficiency, which efficiency is high.