Abstract:

In the distributed storage, the data availability of each distributed storage node is very important. Therefore, the corresponding redundance measures must be taken to ensure the data availability. However, the redundance cost of the traditional redundance strategy based on duplication will greatly increase as data availability improves. Meanwhile, with the increasing of the number of distributed storage nodes, the possibility that some nodes suffer physical attack and theft also increases, which greatly increases the security threat to sensitive data. On the basis of deep research on Raptor code, this paper presents a distributed storage scheme based on Raptor code. And through a lot of experimental tests, a distributed storage system based on wirehair code (open source implementation of RaptorQ) is designed and implemented. The system can not only achieve higher data availability with smaller redundance, but also increase the security of the original data through the coding process.

Key words: distributed storage, wirehair code, redundance, Raptor code