Disk selection method and device for distributed storage cluster and readable storage medium

Abstract

The invention discloses a disk selection method and device for a distributed storage cluster and a computer readable storage medium. The method comprises the steps that a conflict domain used for storing selected object storage equipment and related information of a fault domain where the selected object storage equipment is located is constructed. In the process of sequentially selecting the OSDs downwards from the root node according to the root node information, the fault domain type and the total number of the OSDs to be selected, if the current member is in the conflict domain, selecting other members from the layer where the current member is located; if the current member is not in the conflict domain, the type of the current member is matched with the type of the fault domain, and the object storage device contained in the current member is not selected, member selection is carried out from the next layer of the layer where the current member is located until a target member of the OSD type is selected; and the target member information and the current member information are input into the conflict domain to update the conflict domain, thereby effectively reducing the frequency of invalid disk selection by a data distribution algorithm, improving the disk selection efficiency, and reducing the occupation of CPU resources and the consumption of time.

Description

technical field [0001] The present application relates to the technical field of distributed storage, and in particular to a disk selection method, device and computer-readable storage medium for a distributed storage cluster. Background technique [0002] In a distributed storage system, the stored content is cut according to a fixed size. This piece of fixed-size data is called an object. PG (Placement Group, placement group) is a carrier for placing objects, which is equivalent to the aggregation of multiple objects. A logical concept, PGs and objects are mapped through a consistent hash algorithm, and one PG corresponds to multiple objects. Each PG corresponds to an OSD (Object-based Storage Device, object storage device) through a data distribution algorithm. One PG corresponds to multiple OSDs. An OSD can be simply understood as a physical disk. An OSD can have Multiple PGs, so that the content to be stored is placed on the disk. [0003] In the data distribution alg...

AI Technical Summary

Problems solved by technology

[0004] Related technologies In the process of assigning each OSD member of each PG to each host of the distributed storage system, under a certain randomness, a newly selected OSD member may conflict with an already selected OSD member, and this The situation is especially obvious when the cluster scale is large and the failure domain is the cabinet. The emergence of this situation will undoubtedly increase the number of invalid selections by the data distribution algorithm, and the disk selection based on the data distribution algorithm is inefficient

Images