Practical Research On P2P Technology In Hybrid Cloud Backup System

In our modern information society, data is produced in an explosive way. At the sametime, the importance of data is highlighted increasingly for both individuals andbusinesses. Higher requirements on data availability, scalability and backup and recoveryrate of storage and backup system are expected. The typical cloud backup systemarchitecture consists of three parts: the client, metadata server, storage server. In thisarchitecture, metadata server is easy to become a bottleneck of the whole system, once itcollapses, the whole system is not available. When the distance between client andmetadata sever, storage server is too far and the network environment is not smooth, thebackup and recovery rate will be very low.Therefore, a hybrid cloud backup system architecture is proposed, which is designedto solve problems such as single storage hierarchy, low scalability, low fault tolerance, lowefficiency in existing backup system and backup method. The system consists of threeparts: they are P2P overlay network, super node, and central storage.When P2P technology is applied to backup system, it is necessary for us to researchon some related problems, such as overlay network protocols, data distribution, dataredundancy, node failure detection, replication repair and so on. Because the networktopology in accordance with the Kademlia protocol is simple and efficient, and it is usedin many practical systems, so the system has adopted kademlia overlay network protocols.Data is distributed according to consistent hash algorithm combined with K-bucketsstructure which is a characteristic of Kad. Data redundancy is relied on multiple copies.We use heartbeat mechanism to detect node failure. When we remove an unavailable node,we should repair data copies and modify block position information in time. The testingresults show that the system has good scalability and backup and recovery efficiency.