Research And Implemention Of Distributed File System Based On Moosefs

As the size of the company and the volume of busines increase a lot, the associated development environment and business data also grow exponentially, then the stand-alone server storage space has become to a very prominent problem at present, while accompanied by backup requirements also bring big headache to people. Although the company has a number of large-capacity servers which can be used to store relevant data, but because of these servers are in different rooms, it is very difficult to organize. So this requires us to organize the free spaces of multiple machines to provide more disk storage space, and then to break through the storage limit.There are a large number of enterprises with shared storage server NFS. This method is sufficient when the size is smaller, but with the amount of data grows in combination with the increase of number of visits, the NFS structure is unable to adapt to the situation, performance and user experience will gain big decline. In addition, we also need to ensure that NFS server never down, otherwise all dependent shared data provided in the application is no longer available.Considering the existence of NFS performance bottleneck and single point of failure, a distributed file storage system host began to pop up. In distributed file systems, data is no longer a one-to-many relationship, but many to many relationship. Thus, performance will be increased, and also has high reliability.Therefore, this article begins with the research of traditional distributed systems, comparing the advantages and disadvantages of them, and focuses on analyzing the architecture of moosefs and the realization of its principle; finally we deploy this architecture to our practical system environment to gain more capability and scalability.