Research And Implementation Of Distributed Disaster Recovery System

Disaster recovery system is the most important means to ensure data securityand business continuity when the application system suffers a sudden disaster, suchas earthquakes, fires, floods and other cases. Traditional disaster recovery systemconsists of a production center and a disaster recovery center. If the productioncenter is destroyed, we can make use of the disaster recovery center to restore thedata and business. However, if the disaster recovery system suffers a consecutivedisaster which destroys the production center and disaster recovery center, thesystem can’t ensure the data security any more.In this paper, after studying, analyzing and comparing the advantages anddisadvantages of the mainstream disaster recovery system at home and aboard, basedon the traditional disaster recovery model, we design and implement the disasterrecovery system based on the distributed architecture. In this paper, the main work issummarized as follows:(1)Design and realize a distributed disaster recovery system based on thestructure of1+1+N. The system, called OsnDDR, consists of a production center, acontrol center and N disaster recovery centers, and can ensure the data security underthe conditions of multi-node damage.(2)Propose the PRS algorithm based on RS erasure code to realize the dataredundancy in distributed architecture and reduce the storage resource consumptioncaused by the data redundancy. In addition, RS erasure code is the erasure codewhich performs polynomial operations of field element in Galois Field, so we designthe double-table lookup method to alternative basic arithmetic operations. On theother hand, we propose the parallel implementation method of RS erasure code, andexperiments show that the average performance of the parallel encoding has largelyimproved compared with serial encoding.(3)Build disaster recovery center’s performance model and load model, and theload balancing strategy suitable for OsnDDR to ensure the load balancing of disasterrecovery center.(4)Build the experimental platform and evaluate the OsnDDR system byexperiments. Moreover, we test the influences of OsnDDR to the user’s applications, and the time for encoding by the RS erasure code.(5)Summarize the main works and contributions that are made by researchingand achieving the OsnDDR system. Apart from this, we bring out the expectation forthe future work.