A virtualisation performance evaluation approach for distributed block device

Abstract

With the development of the information age, the volume of data is exploding and the operational and performance requirements of massive data are difficult to balance. The high cost of distributed storage systems with high data reliability and excellent random read and write performance, often consisting of multiple racks, multiple nodes and hundreds of disks, is a thorny issue for researchers. A tool to predict the performance of distributed storage architectures has been built based on a study of the cost of distributed storage architectures. Without hardware, enabling low cost and fast performance prediction. It also provides performance data and latency density to compare the performance of different workloads and architectures more intuitively, providing ideas for optimising the performance of distributed systems.