Research and optimization of distributed file system based on cloud storage

Research and optimization of distributed file system based on cloud storage

University Zhang Ocean

Based on the systematic and comprehensive study and summarization of the development status and characteristics of distributed storage System, this paper focuses on analyzing the advantages and disadvantages of several common distributed storage system architectures, and designs a namenode system architecture with partial equivalence. By increasing some equivalent multiple namenode in the metadata server layer, the architecture changes the single point dependence of the centralized storage system with HDFS, which reduces the latency of concurrent users and the average memory occupancy rate of the metadata server. At the same time, this paper further studies the commonly used load-balancing methods, in view of the disadvantage of the HDFs storage server load Balancing, establishes the disk utilization model and the service blocking rate model, and designs an adaptive feedback load balancing algorithm based on this framework. Through the algorithm performance analysis and experimental simulation, it is further proved that the algorithm designed in this paper is better than the load equalization algorithm in the HDFS system in terms of system performance and load uniformity.

Research and optimization of distributed file system based on cloud storage