Research Of Disaster Recovery System Architecture Based On Virtualization

With the continuous improvement of information degree, the loss of data caused by the disaster of Data Center will lead to company's huge losses. So the construction of Disaster Recovery System is becoming increasingly important for companies. An investigation shows that the main reason that many companies don't construct their Disaster Recovery System is the total cost of ownership (TCO) is too high.This paper improves the utilization of servers in the Disaster Recovery System and decreases TCO by using virtualization on the basis of N+M topology. Realize the non-critical tasks in the Disaster Recovery System to improve the utilization and solve the problem that the Disaster Recovery System is always idle, but this will cause the problem that the Disaster Recovery System cannot provide sufficient resources for all tasks of main system when disaster occurs. Therefore, design and realize an algorithm to determine the key tasks, use this algorithm to choose the key tasks and order these tasks according to their level. Assign resources for high level tasks first to ensure the recovery of the critical tasks.On the base of this algorithm, design a disaster recovery systems based on the virtualization technology, realize the control module and the long-distance recovery module. Generate the key task list in the control module and recover the data according this list in the long-distance recovery module. The final experiment shows that the system reaches the target, at the same time significantly reduces the TCO.