Distributed storage method and system for a block device

Abstract

The invention is applicable to the storage field, particularly relates to distributed storage systems and discloses a distributed storage method and system of a block device. The method comprises defining the storage space of the block device as a logical volume and segmenting the logical volume into N logical blocks in a fixed size; actually storing the logical blocks in a plurality of physical storages, and associating the mapping relations of the logical blocks and a plurality of corresponding physical storage addresses through indexing. The system comprises a data access layer, an index server and a storage server, wherein the data access layer processes a read-write request of the block device on the basis of an iSCSI (Internet small computer system interface) protocol; after analyzing the protocol, the data access layer obtains the actual physical storage position of the logical block involved in the request from the index server; after receiving the request, the index server queries associated indexes to find a corresponding storage server and feeds information back to the data access layer; after receiving information fed back by the index server, the data access layer accesses the storage server to complete subsequent data reading and writing tasks.

Description

technical field [0001] The invention belongs to the technical field of storage, and in particular relates to distributed storage of block devices. Background technique [0002] In distributed storage, there are two main storage methods. One is file-based storage, such as NAS (Network Attached Storage), which is characterized by low cost and easy implementation. In the IO scenario, the efficiency is not high and the performance is poor; the other is block-based storage, such as SAN (Storage Area Network). With the characteristics of high performance, it is very suitable for large-scale deployment as a cloud storage solution. However, because the traditional IP SAN is usually directly connected to the block device on the iSCSI Target side, it is not easy to maintain and expand functions, such as snapshots and clones, and there is no distributed data redundancy to solve the single point of failure problem. Contents of the invention [0003] In order to overcome the deficien...

AI Technical Summary

Problems solved by technology

[0002] In distributed storage, there are two main storage methods. One is file-based storage, such as NAS (Network Attached Storage), which is characterized by low cost and easy implementation. In the IO scenario, the efficiency is not high and the performance is poor; the other is block-based storage, such as SAN (Storage Area Network). With high performance characteristics, it is very suitable for large-scale deployment as a cloud storage solution

However, because the traditional IP SAN is usually directly connected to the block device on the iSCSI Target side, it is not easy to maintain and expand functions, such as snapshots, clones, etc., and there is no distributed data redundancy to solve the single point of failure problem

Images