Management method and device of distributed file system and server

Abstract

The embodiment of the invention provides a management method of a distributed file system, which comprises the following steps of enabling a router to receive information to be processed, wherein theinformation to be processed refers to an operation to be processed, and putting the to-be-processed information into a waiting queue; enabling the scheduler to extract features of the to-be-processedinformation to serve as target features, and select a method for processing the to-be-processed information according to the target features to obtain a preset method; and processing the to-be-processed information by using a preset method to obtain output data; and storing the output data into the configuration distribution network, and updating the data of the configuration distribution networkto obtain updated data; and configuring the distribution network to output the update data except that the client sending the to-be-processed information accesses the update data. Synchronization is carried out when a plurality of users have to access the same data at the same time, so that it is ensured that updating and deleting of the data at the same place can be automatically mapped to otherplaces storing the data.

Description

technical field [0001] The invention relates to the technical field of distributed file systems, in particular to a management method, device and server of a distributed file system. Background technique [0002] In traditional filesystems, locking primitives exist, such as write locks on directories, or file and record locks, to serialize competing requests for the same resource. A file system's stateful implementation must provide similar functionality, and the mechanism for doing so is a per-cabinet queue, which implicitly implements resource serialization. [0003] Traditional serialization mechanisms include the following aspects: spin locks, mutexes, freeze locks, distributed locks, heartbeat mechanisms, and network legal mechanisms. There are various computing models, and the most common model is threaded with resource locks. In distributed computing, this extends to distributed lock managers and heartbeat networks, which extend the single-node concept across multipl...

AI Technical Summary

Problems solved by technology

Various common factors become apparent when analyzing distributed data management systems: implementations are complex, locking models limit performance, locking models limit scale-out, deadlocks are common system downtime, and locking and heartbeat traffic create significant overhead

Images