Metadata read-write method and device of distributed storage system

Abstract

The invention discloses a metadata read-write method and device for a distributed storage system, and the method comprises the steps: responding to a writing request with metadata, randomly writing the metadata into a first storage pool which uses a persistent memory as hardware, and enabling the metadata to directly fall into a disk; in response to the fact that the metadata storage capacity of the first storage pool exceeds a capacity threshold value, continuously determining metadata with the minimum popularity in the first storage pool and copying the metadatainto a second storage pool using the solid state disk as hardware to be directly dropped into a disk until the metadata storage capacity of the first storage pool does not exceed the capacity threshold value; in response to a readrequest for metadata in the second storage pool, reading metadata from the second storage pool and adaptively modifying its popularity; and in response to the fact that the popularity of the metadatain the second storage pool exceeds a popularity threshold, copying the metadata of which the popularity exceeds the popularity threshold into the first storage pool for direct disk falling. Accordingto the method and device, the metadata access performance can be improved, and the overall performance bottleneck of the distributed system is further improved.

Description

technical field [0001] The present invention relates to the field of storage, and more specifically, to a method and device for reading and writing metadata of a distributed storage system. Background technique [0002] In the current distributed NAS (Network Attached Storage) system, data is generally stored on HDD (mechanical hard disk), and metadata is generally stored on SSD (solid state drive). Given that the performance of SSD is much better than that of HDD and the frequency of metadata access is much greater than that of data access, this combination is currently the most common application. At the same time, when data is written to disk, it is used to "write log" first and then "write data" to convert random disk writing into sequential disk writing, which can improve the efficiency of data writing to disk and improve the overall performance. Furthermore, SSD disks are used for grading in the data layer, and the data to be written is written to the SSD layer first,...

AI Technical Summary

Problems solved by technology

Metadata is generally very small. For the system, the read and write of metadata is small file read and write access, and the performance of small file access is very poor. Although SSD is used to store metadata, it is still impossible to change SSD to metadata Access bottleneck problem

[0003] There is currently no effective solution to the problem that metadata access becomes a system bottleneck in the prior art

Images