Granted Patent B2

US 12,265,455 · App. 17/452,787 · Granted Apr 1, 2025

Inventors: Guang Han Sui (Beijing, CN); Wei Ge (Beijing, CN); Lan Zhe Liu (Beijing, CN); Zhang Li Ping (Beijing, CN); Er Tao Zhao (Beijing, CN)

Assignee: International Business Machines Corporation

G06F11/203G06F9/461G06F9/485G06F9/4856G06F9/5038G06F11/0709G06F2209/5021G06F2209/509

Abstract

Claims (31)

1. A computer-implemented method for task failover in a cloud environment, wherein the cloud environment includes a plurality of reclaimable nodes, comprising:

monitoring, by one or more processing units, if any node of the plurality of reclaimable nodes having one or more child nodes is to be reclaimed for executing other tasks, a reclaimable node executing in a spot instance in the cloud environment and being reclaimable at any time without delay for data storing;

determining, by one or more processing units, whether a task of the existing tasks on any node of the reclaimable nodes having one or more child nodes is recoverable, the task defined as recoverable if the task has been executed on the one or more child nodes for a period higher than a threshold and an impact will occur if associated task execution in the one or more child nodes is killed, the impact related to wasted calculations associated with task execution in the one or more child nodes;

storing, by one or more processing units, data of the recoverable task on the reclaimable node remotely in a stable node;

notifying, by one or more processing units, at least one associated task executed by at least one child node associated with a reclaimed node in the cloud environment to wait without abandoning the at least one associated task which results in a waste of calculation resources and delay of response;

connecting, by one or more processing units, the child node of the reclaimed node to the stable node which will not be reclaimed to interrupt task execution; and

continuing execution of the recoverable task and the at least one associated task on the stable node.

2. The method of claim 1 , wherein the step of determining whether a task is recoverable is further based on at least one of the following factors associated with the task on the to-be-reclaimed node: task execution time, a task execution percentage, a task execution result size, a number of child nodes, and a number of times of being re-executed.

3. The method of claim 1 , wherein monitoring if any node of the plurality of reclaimable nodes is to be reclaimed, determining whether a task of the existing tasks on any node of the plurality of reclaimable nodes is recoverable, and storing data of the recoverable task remotely in the stable node occur in parallel.

4. The method of claim 1 , further comprises:

responsive to the at least one node being reclaimed and the task being unrecoverable, exiting, by one or more processing units, the task execution of the unrecoverable task on the reclaimed node and the task execution associated with the unrecoverable task on the child node of the reclaimed node.

5. A computer program product for task failover in a cloud environment, wherein the cloud environment includes a plurality of reclaimable nodes, the computer program product comprising:

a computer readable storage medium having program instructions embodied therewith, the program instructions being executable by a computer to cause the computer to perform a method comprising:

monitoring if any node of the plurality of reclaimable nodes having one or more child nodes is to be reclaimed for executing other tasks, a reclaimable node executing in a spot instance in the cloud environment and being reclaimable at any time without delay for data storing;

determining whether a task of the existing tasks on any node of the reclaimable nodes having one or more child nodes is recoverable, the task defined as recoverable if the task has been executed on the one or more child nodes for a period higher than a threshold and an impact will occur if associated task execution in one or more child nodes is killed, the impact related to wasted calculations associated with task execution in the one or more child nodes;

storing data of the recoverable task on the reclaimable node remotely in a stable node;

notifying at least one associated task executed by at least one child node associated with a reclaimed node in the cloud environment to wait without abandoning the at least one associated task which results in a waste of calculation resources and delay of response;

connecting the child node of the reclaimed node to the stable node which will not be reclaimed to interrupt task execution; and

continuing execution of the recoverable task and the at least one associated task on the stable node.

6. The computer program product of claim 1 , wherein the step of determining whether a task is recoverable is further based on at least one of the following factors associated with the task on the reclaimed node: task execution time, a task execution percentage, a task execution result size, a number of child nodes, and a number of times of being re-executed.

7. A computer system for task failover in a cloud environment, wherein the cloud environment includes a plurality of reclaimable nodes, comprising:

one or more processors;

a memory coupled to at least one of the processors; and

a set of computer program instructions stored in the memory and executed by at least one of the processors to perform a method comprising:

monitoring if any node of the plurality of reclaimable nodes having one or more child nodes is to be reclaimed for executing other tasks, a reclaimable node executing in a spot instance in the cloud environment and being reclaimable at any time without delay for data storing;

determining whether a task of the existing tasks on any node of the reclaimable nodes having one or more child nodes is recoverable, the task defined as recoverable if the task has been executed on the one or more child nodes for a period higher than a threshold and an impact will occur if associated task execution in the one or more child nodes is killed, the impact related to wasted calculations associated with task execution in the one or more child nodes;

storing data of the recoverable task on the reclaimable node remotely in a stable node;

notifying at least one associated task executed by at least one child node associated with a reclaimed node in the cloud environment to wait without abandoning the at least one associated task which results in a waste of calculation resources and delay of response;

connecting, by one or more processing units, the child node of the reclaimed node to the stable node which will not be reclaimed to interrupt task execution; and

continuing execution of the recoverable task and the at least one associated task on the stable node.

8. The computer system of claim 7 , wherein the step of determining whether a task is recoverable is further based on at least one of the following factors associated with the task on the reclaimed node: task execution time, a task execution percentage, a task execution result size, a number of child nodes, and a number of times of being re-executed.

Assignments (1)

Continuity (1)

References Cited (47)