Elastic Execution of Machine Learning Workloads Using Application Based Profiling

Patent metadata

Field	Value
Publication number	US-2021191759-A1
Application number	US-201916724613-A
Country	US
Kind code	A1
Filing date	Dec 23, 2019
Priority date	Dec 23, 2019
Publication date	Jun 24, 2021
Grant date	—

Abstract

Embodiments relate to a system, program product, and method for supporting elastic execution of a machine learning (ML) workload using application based profiling. A joint profile comprised of both ML application execution and resource usage data is generated. One or more feature(s) and signature(s) from the joint profile are identified, and a ML execution model for ML application execution and resource usage is built. The ML execution model leverages the feature(s) and signature(s) and is applied to provide one or more directives to subsequent application execution. The application of the ML execution model supports and enables the ML execution to elastically allocate and request one or more resources from a resource management component, with the elastic allocation supporting application execution.

First claim

What is claimed is:

1 . A computer system comprising:
a processing unit operatively coupled to memory; an artificial intelligence (AI) platform, in communication with the processing unit, having tools to elastically execute one or more machine learning workloads using application based profiling, the tools comprising:
a profile manager to generate a joint profile comprised of machine learning (ML) application execution and resource usage;
a machine learning manager to identify one or more features and one or more signatures from the generated joint profile and build a ML execution model for ML application execution performance and resource usage, the ML execution model to leverage the identified one or more features and one or more signatures; and
a director to apply the ML execution model and provide one or more directives to subsequent application execution, including the ML execution model to elastically allocate and request one or more resources from a resource management component to support application execution.

2 . The computer system of claim 1 , wherein elastic allocation includes the ML execution model to execute a resource allocation action with respect to resource availability and support for application processing, the resource allocation action to reduce or expand any resource, and further wherein the one or more reduced or expanded resource having the same or different attributes.

3 . The computer system of claim 2 , further comprising the director to change computing allocation across iterations of one or more ML applications, including invoke the resource allocation action based on an application execution pattern and a resource usage pattern.

4 . The computer system of claim 1 , wherein the joint profile generation further comprises the profile manager to monitor and collect resource usage data on one or more ML routines, and use the monitored data to predict allocation for a future ML application.

5 . The computer system of claim 1 , wherein the profile manager utilizes one or more call-back functions to profile application execution with one or more application relevant parameters and profile resource usage information.

6 . The computer system of claim 5 , wherein the generated joint profile is collected for a distributed ML application in a distributed computing system.

7 . The computer system of claim 1 , wherein identification of one or more features and one or more signatures for joint profile generation and building of execution model is derived with user input.

8 . A computer program product to support elastic execution of a machine learning workload using application based profiling, the computer program product comprising a computer readable storage medium having program code embodied therewith, the program code executable by a processor to:
generate a joint profile comprised of machine learning (ML) application execution and resource usage; identify one or more features and one or more signatures from the generated joint profile and build a ML execution model for ML application execution performance and resource usage, the ML execution model to leverage the identified one or more features and one or more signatures; and apply the ML execution model and provide one or more directives to subsequent application execution, including elastically allocating and requesting one or more resources from a resource management component to support application execution.

9 . The computer program product of claim 8 , wherein elastic allocation includes the program code to execute a resource allocation action with respect to resource availability and support for application processing, the resource allocation action to reduce or expand any resource, and further wherein the reduced or expanded resource having the same or different attributes.

10 . The computer program product of claim 9 , further comprising the processor to change computing resource allocation across iterations of one or more ML applications, including invoke the resource allocation action based on an application execution pattern and a resource usage pattern.

11 . The computer program product of claim 8 , wherein the joint profile generation further comprises the program code to monitor and collect resource usage data on one or more ML routine, and use the monitored data to predict allocation for a future ML application.

12 . The computer program product of claim 8 , wherein the program code to generate the joint profile utilizes one or more call-back functions to profile application execution with one or more application relevant parameters and profile resource usage information.

13 . The computer program product of claim 12 , wherein the generated joint profile is collected for a distributed ML application in a distributed computing system.

14 . The computer program product of claim 8 , wherein identification of one or more features and one or more signatures for joint profile generation and building of execution model can be derived with user input.

15 . A method comprising:
generating a joint profile comprised of machine learning (ML) application execution and resource usage; identifying one or more features and one or more signatures from the generated joint profile and building a ML execution model for ML application execution performance and resource usage, the ML execution model leveraging the identified one or more features and one or more signatures; applying the ML execution model and providing a directive to subsequent application execution, including elastically allocating and requesting one or more resources from a resource management component to support application execution.

16 . The method of claim 15 wherein elastic allocation includes execution of a resource allocation action with respect to resource availability and support for application processing, the resource allocation action to reduce or expand any resources, and further wherein the reduced or expanded resource having the same or different attributes.

17 . The method of claim 16 , further comprising changing computing resource allocation across one or more iterations of one or more ML applications, including invoking the resource allocation action based on an application execution pattern and a resource usage pattern.

18 . The method of claim 15 , wherein generating the joint profile further comprises monitoring and collecting resource usage data on one or more ML routines, and using the monitored data to predict allocation for a future ML application.

19 . The method of claim 15 , wherein generating the joint profile utilizes one or more call-back functions to profile application execution with one or more application relevant parameters and profile resource usage information.

20 . The method of claim 19 , wherein the generated joint profile is collected for a distributed ML application in a distributed computing system.

Assignees

• IBM

Inventors

• Fong Liana
• Seelam Seetharami R
• Venkataraman Ganesh
• Saha Debashish
• Oum Punleuk
• Verma Archit
• Reddy Prabhat Maddikunta

Classifications

• G06N20/00

  Machine learning · CPC title

• G06N5/02

  Knowledge representation; Symbolic representation · CPC title

• G06N3/08

  Learning methods · CPC title

• G06F9/5072 Primary

  Grid computing · CPC title

• G06F2209/482

  Application · CPC title