Workflow optimization and re-distribution

Patent metadata

Field	Value
Publication number	US-12566644-B2
Application number	US-202217809570-A
Country	US
Kind code	B2
Filing date	Jun 29, 2022
Priority date	Jun 29, 2022
Publication date	Mar 3, 2026
Grant date	Mar 3, 2026

Abstract

Workflow optimization and re-distribution involves a target task set that is determined from a plurality of tasks in a workflow and based on a data flow of the workflow. The target task set is re-distributed to obtain at least one re-distributed task set. Each re-distributed task set indicates a respective data path different from a data path of the target task set and has a same function as the target task set. A simulated data flow for the at least one re-distributed task set is obtained by simulating data flowing in at least one data path indicated by the at least one re-distributed task set. An optimized task set for the target task set is determined based on the simulated data flow for the at least one re-distributed task set.

First claim

What is claimed is:

1 . A computer-implemented method comprising:
determining a target task set from a plurality of tasks included in a workflow based on a data flow of the workflow; determining similarities among the plurality of tasks based on the data flow of the workflow, wherein a task similarity is calculated by leveraging input/output pattern data from an intercepted request from the target task set; selecting the target task set with a potential bottleneck of the workflow, wherein the potential bottleneck comprises intensive input/output pattern data of the target task set; re-distributing the target task set before deployment to avoid the potential bottleneck and to obtain at least one re-distributed task set for the target task set, wherein the at least one re-distributed task set indicates a respective data path different from a data path of the target task set, and wherein the at least one re-distributed task set has a same function as the target task set; determining an optimized task set for the target task set based on data flow simulation and analysis; and executing the workflow by sharing the optimized task set without a potential security risk.

2 . The method of claim 1 wherein determining the target task set further includes:
grouping the plurality of tasks of the workflow into a plurality of task sets based on the data flow of the workflow and a knowledge base; and selecting the target task set from the plurality of task sets.

3 . The method of claim 2 , wherein selecting the target task set further includes:
determining the task set as the target task set based, at least in part, upon a determination that a time duration for completing the task exceeds a time threshold and/or a determination that a resource consumption for completing the task exceeds a consumption threshold.

4 . The method of claim 2 , wherein grouping the plurality of tasks into the plurality of task sets further includes:
grouping the plurality of tasks into the plurality of task sets based, at least in part, upon operators of the plurality of tasks and/or functions of the plurality of tasks.

5 . The method of claim 2 , wherein selecting the target task set further includes:
determining similarities among the plurality of task sets based on the data flow of the workflow; and in accordance with a determination that a similarity between a first task set and a second task set exceeds a threshold, determining the first task set or the second task set to be the target task set.

6 . The method of claim 5 , wherein determining the similarities includes:
for each task set of the plurality of task sets:
obtaining input data and output data of the task set by intercepting a request to the task set;
aggregating a data metric for the task set based at least in part on the input data and output data of the task set;
obtaining data flow timeline information of the task set; and
determining the similarities based on the data metrics and data flow timeline information of the plurality of task sets.

7 . The method of claim 6 , wherein aggregating the data metric for the task set includes:
aggregating the data metric based on the input data and output data of the task set, pre-conditions and post-conditions of the task set and task implementations of the task set.

8 . The method of claim 1 , wherein obtaining the simulated data flow for the at least one re-distributed task set further includes:
generating simulated input data for the at least one re-distributed task set; obtaining simulated output data by inputting the simulated input data to the at least one data path indicated by the at least one re-distributed task set; and obtaining the simulated data flow by collecting the simulated input data and the simulated output data along with data flow timeline information and resource consumption information for the at least one re-distributed task set.

9 . The method of claim 1 , wherein determining the optimized task set further includes:
determining the at least one re-distributed task set to be the optimized task set based, at least in part, upon a determination that a time duration for completing the at least one re-distributed task set is below a time threshold and/or a determination that resource consumption for completing the at least one re-distributed task is below a consumption threshold.

10 . The method of claim 1 , wherein the data flow of the workflow includes: (i) input data of tasks in the workflow; (ii) output data of tasks in the workflow; (iii) pre-conditions of tasks in the workflow; (iv) post-conditions of tasks in the workflow; (v) timeline information of tasks in the workflow; and/or (vi) resource consumption information of tasks in the workflow.

11 . The method of claim 1 , further comprising:
determining a further similarity between a further task set of a further workflow and the target task set; and in accordance with a determination that the further similarity exceeds a threshold, re-distributing the further task set of the further workflow based on the optimized task set.

12 . A computer system comprising:
a processing unit; and a memory coupled to the processing unit and storing instructions thereon, the instructions, when executed by the processing unit, performing operations including:
determining a target task set from a plurality of tasks included in a workflow based on a data flow of the workflow;
determining similarities among the plurality of tasks based on the data flow of the workflow, wherein a task similarity is calculated by leveraging input/output pattern data from an intercepted request from the target task set;
selecting the target task set with a potential bottleneck of the workflow, wherein the potential bottleneck comprises intensive input/output pattern data of the target task set;
re-distributing the target task set before deployment to avoid the potential bottleneck and to obtain at least one re-distributed task set for the target task set, wherein the at least one re-distributed task set indicates a respective data path different from a data path of the target task set, and wherein the at least one re-distributed task set has a same function as the target task set;
determining an optimized task set for the target task set based on data flow simulation and analysis; and
executing the workflow by sharing the optimized task set without a potential security risk.

13 . The computer system of claim 12 further comprising:
grouping the plurality of tasks of the workflow into a plurality of task sets based on the data flow of the workflow and a knowledge base; and
selecting the target task set from the plurality of task sets.

14 . The computer system of claim 13 wherein selecting the target task set further includes:
determining the task set as the target task set based, at least in part, upon a determination that a time duration for completing a task of the plurality of task sets exceeds a time threshold and/or a determination that resource computation for completing the task set exceeds a consumption threshold.

15 . The computer system of claim 13 , wherein grouping the plurality of tasks into the plurality of task sets further includes:
grouping the plurality of tasks into the plurality of task sets based, at least in part, upon operators of the plurality of tasks and/or functions of the plurality of tasks.

16 . The computer system of claim 13 , wherein selecting the target task set further includes:
determining similarities among the plurality of task sets based on the data flow of the workfl

Assignees

• IBM

Inventors

• Jiang peng hui
• Guan Zhi Li
• Su Jun
• Yang Kun
• Diao Yun
• Su Shi

Classifications

• G06F9/5066

  Algorithms for mapping a plurality of inter-dependent sub-tasks onto a plurality of physical CPUs (mappping at compile time, see G06F8/451) · CPC title

• G06N5/02

  Knowledge representation; Symbolic representation · CPC title

• G06F9/5038

  considering the execution order of a plurality of tasks, e.g. taking priority or time dependency constraints into consideration (scheduling strategies G06F9/4881 and subgroups) · CPC title

• G06F9/5072

  Grid computing · CPC title

• G06F9/5088 Primary

  involving task migration · CPC title