With the development of networking technology and storage technology, networking storage technology was born and made up the shortcomings of simple storage. Data backup and disaster recovery are the key issues in the study of networking storage. Three-site Disaster Recovery Scheme combines local subsystem providing high availability though automatic swap and remote subsystem which provides high reliability, thus providing enterprises with the most secure and reliable storage services. Compiled with the traditional Fibre-based storage and disater recovery system, IP networking storage technology has the advantage of low cost, good performance and high scalability.This paper focuses on the study of three-site disaster recovery scheme for the need of median-scale companies. An iSCSI-based Cascading Remote Copy System is designed and implemented for the scheme. The implementation model is done on personal computers. At first we implement a system with basic functions of data backup and disaster recovery. The system is composed of two cascading parts. One part is local subsystem with the functionality of real-time data copy and automatic swap. The other part is remote subsystem with remote-site data recovery function. For the shortcomings and limits such as low performance of local subsystem and poor coordination of the two subsystems, we optimized the basic functional system. We introduced concurrent technology with data integrity control into the synchronous remote copy system to enhance the performance of local subsystem, and we put the metadata record methods into the asynchronous remote copy system to improve the coordination of the two subsystems when the networking circumstances of the two subsystems differ a lot.System performance evaluation and functional tests show that the optimized system has much better read/write performance, closer to that of the original iSCSI system. It also has good functionality of storage, backup and disaster recovery, performing normally even when the two subsystems are working in different networking circumstances. The system provides median-scale enterprises with a good model of implementing the three-site disaster recovery sheme with high performance-cost ratio. |