Data storage method and apparatus

Patent metadata

Field	Value
Publication number	US-11036535-B2
Application number	US-201816188951-A
Country	US
Kind code	B2
Filing date	Nov 13, 2018
Priority date	Nov 21, 2016
Publication date	Jun 15, 2021
Grant date	Jun 15, 2021

Abstract

A data storage method and a physical server are provided. M virtual machines are deployed on a plurality of physical servers. The M virtual machines are respectively deployed as M data nodes in a distributed storage system. A metadata node in the distributed storage system receives a data storage request of a client, and determines identifiers of N virtual machines from the M virtual machines based on stored grouping information. The grouping information records a mapping relationship between a plurality of anti-affinity groups and identifiers of the M virtual machines.

First claim

What is claimed is:

1. A method, comprising:
receiving, by a metadata node (MDS) in a distributed storage system, a data storage request of a client, wherein the distributed storage system comprises M data nodes, the M data nodes are deployed on M virtual machines, and the M is a positive integer greater than or equal to 2;
determining, by the MDS, that a quantity of storage copies of data specified in the data storage request is N, wherein the N is a positive integer greater than or equal to 1;
determining, by the MDS, identifiers of N virtual machines of the M virtual machines, wherein the identifiers of the N virtual machines are determined based on grouping information, a mapping relationship between a plurality of anti-affinity groups and identifiers of the M virtual machines is recorded in the grouping information, wherein the N virtual machines belong to at least one anti-affinity group of the plurality of anti-affinity groups, and wherein virtual machines in each anti-affinity group of the plurality of anti-affinity groups have anti-affinity with each other; and
sending, by the MDS, response information of the data storage request to the client, wherein the response information comprises the identifiers of the N virtual machines, and the response information instructs the client to store, in the N virtual machines, N copies of the data specified in the data storage request.

2. The method according to claim 1 , wherein the determining the identifiers of the N virtual machines, wherein the identifiers of the N virtual machines are determined based on the grouping information, comprises:
when a quantity of the virtual machines in the each anti-affinity group of the plurality of anti-affinity groups is greater than or equal to the N, determining, by the MDS, the identifiers of the N virtual machines from one of the plurality of anti-affinity groups; or
when the quantity of the virtual machines in the each anti-affinity group of the plurality of anti-affinity groups is less than the N, determining, by the MDS, the identifiers of the N virtual machines from at least two anti-affinity groups of the plurality of anti-affinity groups.

3. The method according to claim 2 , wherein the determining, by the MDS, the identifiers of the N virtual machines from the at least two anti-affinity groups of the plurality of anti-affinity groups comprises:
determining, by the MDS, the identifiers of the N virtual machines from the at least two anti-affinity groups based on affinity.

4. The method according to claim 3 , wherein the M virtual machines belong to a plurality of affinity groups, virtual machines in each affinity group of the plurality of affinity groups have affinity with each other, and a mapping relationship between the plurality of affinity groups and the identifiers of the M virtual machines is recorded in the grouping information.

5. The method according to claim 4 , wherein the virtual machines in the each affinity group are deployed on a same physical server.

6. The method according to claim 1 , further comprising:
receiving, by the MDS, the identifiers of the M virtual machines sent by a virtual machine manager;
grouping, by the MDS, the M virtual machines into the plurality of anti-affinity groups, and recording, in the grouping information, the mapping relationship between the plurality of anti-affinity groups and the identifiers of the M virtual machines; and
sending, by the MDS, the grouping information to the virtual machine manager.

7. The method according to claim 1 , wherein any two of the virtual machines in the each anti-affinity group are deployed on different physical servers.

8. A method, comprising:
sending, by a virtual machine manager, identifiers of M virtual machines to a metadata node (MDS) in a distributed storage system, wherein the distributed storage system comprises M data nodes, the M data nodes are deployed on the M virtual machines, and the M is a positive integer greater than or equal to 2;
receiving, by the virtual machine manager, grouping information sent by the MDS, wherein a mapping relationship between a plurality of anti-affinity groups and the identifiers of the M virtual machines is recorded in the grouping information, and virtual machines in each anti-affinity group of the plurality of anti-affinity groups have anti-affinity with each other; and
deploying, by the virtual machine manager, the M virtual machines based on a comparison between the grouping information and an actual distribution status of the M virtual machines, wherein, after the deploying, any two of the virtual machines in the each anti-affinity group of the plurality of anti-affinity groups are deployed on different physical servers, the deploying comprising:
determining, by the virtual machine manager, that the actual distribution status of the M virtual machines is inconsistent with the grouping information, wherein a first virtual machine and a second virtual machine of a first anti-affinity group of the plurality of anti-affinity groups are on a first physical server of the different physical servers; and
based on the actual distribution status and the grouping information, migrating, by the virtual machine manager, the second virtual machine from the first physical server to a second physical server of the different physical servers.

9. The method according to claim 8 , wherein the M virtual machines belong to a plurality of affinity groups, virtual machines in each affinity group of the plurality of affinity groups have affinity with each other, and a mapping relationship between the plurality of affinity groups and the identifiers of the M virtual machines is recorded in the grouping information.

10. The method according to claim 9 , wherein the virtual machines in the each affinity group are deployed on a same physical server.

11. A physical server, comprising:
a non-transitory memory storing instructions;
at least one processor, wherein the at least one processor is configured to execute the instructions in the non-transitory memory to:
receive, for a metadata node (MDS) in a distributed storage system, a data storage request of a client, wherein the distributed storage system comprises M data nodes, the M data nodes are deployed on M virtual machines, and the M is a positive integer greater than or equal to 2;
determine, for the MDS, that a quantity of storage copies of data specified in the data storage request is N, wherein N is a positive integer greater than or equal to 1;
determine, for the MDS, identifiers of N virtual machines of the M virtual machines,

wherein the identifiers of the N virtual machines are determined based on grouping information, wherein a mapping relationship between a plurality of anti-affinity groups and identifiers of the M virtual machines is recorded in the grouping information, the N virtual machines corresponding to the identifiers of the N virtual machines belong to at least one anti-affinity group of the plurality of anti-affinity groups, and wherein virtual machines in each anti-affinity group of the plurality of anti-affinity groups have anti-affinity with each other; and
send, for the MDS, response information of the data storage request to the client, wherein the response information comprises the identifiers of the N virtual machines, and the response information instructs the client to store, in the N virtual machines, N copies of the data specified in the data storage request.

12. The physical server according to claim 11 , wherein the at least one processor is configured to execute the instructions to:
determine, for the MDS, the identifiers of the N virtual machines from one of the plurality of an

Assignees

• Huawei Tech Co Ltd

Inventors

• Zhong Yong
• Lin Ming
• Peng Ruilin
• Zhao Jun

Classifications

• G06F3/0614 Primary

  Improving the reliability of storage systems · CPC title

• G06F3/0604

  Improving or facilitating administration, e.g. storage management · CPC title

• G06F3/065

  Replication mechanisms · CPC title

• G06F11/2094

  Redundant storage or storage space (G06F11/2056 takes precedence) · CPC title

• G06F2009/45562

  Creating, deleting, cloning virtual machine instances · CPC title