Method for creating consistency snapshot for distributed application, apparatus, and distributed system

Patent metadata

Field	Value
Publication number	US-11080150-B2
Application number	US-201916584862-A
Country	US
Kind code	B2
Filing date	Sep 26, 2019
Priority date	Mar 23, 2018
Publication date	Aug 3, 2021
Grant date	Aug 3, 2021

Abstract

A distributed system for creating a consistency snapshot for a distributed application includes a control node and multiple execution nodes including a source execution node, an intermediate execution node, and a leaf execution node. The source execution node receives a snapshot event, and generates a single-point snapshot of an application instance associated with the source execution node. Each of the intermediate execution node and the leaf execution node sequentially receives snapshot events separately sent by two parent nodes of the respective nodes, generates a single-point snapshot of an application instance associated with the respective nodes after receiving the first snapshot event, records a log of a task when executing the task triggered by another parent node different from a parent node sending the first snapshot event, and stops recording a log of a task triggered each time one snapshot event sent by a parent node is received subsequently.

First claim

What is claimed is:

1. A distributed system for creating a consistent snapshot, the distributed system comprising:
a control node; and
a plurality of execution nodes comprising a source execution node, an intermediate execution node, and a leaf execution node, each execution node being associated with an application instance of a distributed application, wherein:
the control node creates the consistency snapshot for the distributed application by using the plurality of execution nodes and is configured to send a snapshot event to the source execution node;
the source execution node is configured to:
receive the snapshot event sent by the control node;
generate a single-point snapshot of an application instance associated with the source execution node; and
send a snapshot event to a child node of the source execution node;

the intermediate execution node is configured to:
sequentially receive snapshot events separately sent by at least two parent nodes of the intermediate execution node;
generate a single-point snapshot of an application instance associated with the intermediate execution node after receiving a first snapshot event;
record a log of a task when executing the task triggered by another parent node of the intermediate execution node different from a parent node sending the first snapshot event;
stop recording a log of a task triggered by the parent node each time one snapshot event sent by a parent node of the intermediate execution node is received subsequently; and
send a snapshot event to a child node of the intermediate execution node after confirming that snapshot events sent by all parent nodes of the intermediate execution node are received; and

the leaf execution node is configured to:
sequentially receive snapshot events separately sent by at least two parent nodes of the leaf execution node;
generate a single-point snapshot of an application instance associated with the leaf execution node after receiving a first snapshot event;
record a log of a task when executing the task triggered by another parent node of the leaf execution node different from a parent node sending the first snapshot event; and
stop recording a log of a task triggered by a sender of a currently received snapshot event each time one snapshot event sent by a parent node of the leaf execution node is received subsequently.

2. The system according to claim 1 , wherein the control node is further configured to:
obtain an attribute of each application instance comprised in the distributed application; and
determine, based on the obtained attribute of the application instance, a connection relationship between a plurality of application instances comprised in the distributed application; and
send the connection relationship to the plurality of execution nodes.

3. The system according to claim 2 , wherein the attribute of each application instance comprises an instance type, a child node type, and a source node mark.

4. The system according to claim 2 , wherein the control node is further configured to:
receive an application protection plan creation message, the application protection plan creation message carrying an identifier of the distributed application;
obtain, based on the identifier of the distributed application, the attribute of each application instance comprised in the distributed application; and
store the connection relationship in a memory.

5. The system according to claim 1 , wherein the source execution node is further configured to return a snapshot complete event to the control node after generating the single-point snapshot;
each of the intermediate execution node and the leaf execution node is further configured to return a snapshot complete event to the control node after generating the single-point snapshot and receiving snapshot events sent by all parent nodes; and
the control node is further configured to determine, after receiving snapshot complete events returned by the execution nodes associated with all application instances of the distributed application, that creation of the consistency snapshot of the distributed application is complete.

6. The system according to claim 1 , wherein the control node is further configured to instruct the plurality of execution nodes to perform a snapshot restoration operation;
the source execution node is further configured to restore data of the application instance associated with the source execution node based on the single-point snapshot of the source application instance;
the intermediate execution node is further configured to:
restore, based on the single-point snapshot of the application instance associated with the intermediate execution node, data of the application instance associated with the intermediate execution node; and
execute the task comprised in the log; and

the leaf execution node is further configured to:
restore, based on the single-point snapshot of the application instance associated with the leaf execution node, data of the application instance associated with the leaf execution node; and
execute the task comprised in the log.

7. A method for creating a consistency snapshot for a distributed application in a distributed system, the distributed system comprising a control node and a plurality of execution nodes, each execution node being associated with an application instance of the distributed application, the plurality of execution nodes comprising a source execution node, an intermediate execution node, and a leaf execution node, the method comprising:
sending, by the control node, a snapshot event to the source execution node;
receiving, by the source execution node, the snapshot event sent by the control node;
generating, by the source execution node, a single-point snapshot of an application instance associated with the source execution node; and
sending, by the source execution node, a snapshot event to a child node of the source execution node;
sequentially receiving, by the intermediate execution node, snapshot events separately sent by at least two parent nodes of the intermediate execution node;
generating, by the intermediate execution node, a single-point snapshot of an application instance associated with the intermediate execution node after receiving a first snapshot event;
recording, by the intermediate execution node, a log of a task when executing the task triggered by another parent node of the intermediate execution node different from a parent node sending the first snapshot event;
stopping, by the intermediate execution node, recording a log of a task triggered by the parent node each time one snapshot event sent by a parent node of the intermediate execution node is received subsequently; and
sending, by the intermediate execution node, a snapshot event to a child node of the intermediate execution node after confirming that snapshot events sent by all parent nodes of the intermediate execution node are received; and
sequentially receiving, by the leaf execution node, snapshot events separately sent by at least two parent nodes of the leaf execution node;
generating, by the leaf execution node, a single-point snapshot of an application instance associated with the leaf execution node after receiving a first snapshot event;
recording, by the leaf execution node, a log of a task when executing the task triggered by another parent node of the leaf execution node different from a parent node sending the first snapshot event; and
stopping, by the leaf execution node, recording a log of a task triggered by the sender of the currently received snapshot event each time one snapshot event sent by a parent node of the leaf execution node is received subsequently.

8. The method according to claim 7

Assignees

• Huawei Tech Co Ltd

Inventors

• Yin Wei
• Luo Bin

Classifications

• G06F16/128 Primary

  Details of file system snapshots on the file-level, e.g. snapshot creation, administration, deletion (error detection or correction of the data by redundancy in operations or in hardware G06F11/14, G06F11/16) · CPC title

• G06F11/1451

  by selection of backup contents · CPC title

• H04L67/1097

  for distributed storage of data in networks, e.g. transport arrangements for network file system [NFS], storage area networks [SAN] or network attached storage [NAS] · CPC title

• G06F9/461

  Saving or restoring of program or task context · CPC title

• G06F11/1464

  for networked environments · CPC title