Model-based resource allocation for an information handling system
User responsiveness on an information handling system may be improved by classifying an application based on its importance and/or relevance for an individual user with the goal of prioritizing resource allocation to improve responsiveness and performance of applications. The classification may include analyzing telemetry data to determine the most important applications for a user, such as by determining an application's importance and/or relevance to a particular user, and determine the resource utilization of that application from a macro perspective. After classification, changing characteristics of an application may be monitored and used to dynamically allocate system resources to the application during runtime. In this manner, priority on resource allocations for certain resources may be adapted to fit the user and the application, and adapt to the changing requirements and scenarios. The determination of application importance and/or relevance and subsequent adaptation of system resource allocation may be performed using a model-based algorithm.
1. A method, comprising:
executing, by an information handling system, an application for a user, wherein executing an application for a user comprises:
determining, by the information handling system via a machine learning model, a priority of the application to the user, wherein the machine learning model determines the priority based, at least in part, on telemetry data, wherein the telemetry data describe user interactions performed with the application, tasks performed by the application, and resources utilized by the application;
subsequent to the determination of the priority of the application, placing the application on the whitelist;
based on the application is on the application whitelist, allocating system resources of the information handling system to the application by performing steps comprising:
determining, by the information handling system, a macro classification for the application based on a resource utilization associated with the application from the application whitelist using a second model, wherein the second model is trained, at least in part, on the telemetry data and characteristic information associated with resource utilization for tasks performed by the application;
determining, by the information handling system, system resources for functions being performed by the application using a third model, wherein the third model is trained, at least in part, on the telemetry data and real-time metrics associated with resources for tasks performed by the application;
allocating, by the information handling system, the system resources based, at least in part, on the macro classification and the system resources for tasks performed by the application;
updating, by the information handling system, usage of the allocated system resources of the information handling system by the application;
updating, by the information handling system, the macro classification for the application whitelist based, at least in part, on the updated usage of the system resources using the second model;
updating, by the information handling system, the resource utilization for tasks performed by the application based, at least in part, on the updated usage of the system resources using the third model; and
reallocating, by the information handling system, the system resources based, at least in part, on the updated macro classification and the updated resource utilization for tasks performed by the application.
2. The method of claim 1 , wherein updating the application whitelist comprises updating the macro classification for resource utilization associated with the application on the application whitelist.
3. The method of claim 1 , wherein the third model is trained to predict resource usage by the application based on a function being performed by the application.
4. An information handling system, comprising:
a processor;
a memory coupled to the processor,
wherein the processor is configured to perform operations comprising:
executing an application for a user, wherein executing an application for a user comprises:
determining, via a machine learning model, a priority of the application to the user, wherein the machine learning model determines the priority based, at least in part, on telemetry data, wherein the telemetry data describe user interactions performed with the application, tasks performed by the application, and resources utilized by the application;
subsequent to the determination of the priority of the application, placing the application on the whitelist;
based on the application being placed on the application whitelist allocating system resources of the information handling system to the application by performing steps comprising:
determining a macro classification for the application based on a resource utilization associated with the application from the application whitelist using a second model, wherein the second model is trained, at least in part, on the telemetry data and characteristic information associated with resource utilization for tasks performed by the application;
determining a system resources for functions being performed by the application using a third model, wherein the third model is trained, at least in part, on the telemetry data and real-time metrics associated with resources for tasks performed by the application;
allocating the system resources based, at least in part, on the macro classification and the system resources for tasks performed by the application;
updating usage of the allocated system resources of the information handling system by the application;
updating the macro classification for the application whitelist based, at least in part, on the updated usage of the system resources using the second model;
updating the resource utilization for tasks performed by the application based, at least in part, on the updated usage of the system resources using the third model; and
reallocating the system resources based, at least in part, on the updated macro classification and the updated resource utilization for tasks performed by the application.
5. The information handling system of claim 4 , wherein updating the application whitelist comprises updating the utilization associated with the application on the application whitelist.
6. The information handling system of claim 4 , wherein the third model is trained to predict resource usage by the application based on a function being performed by the application.
7. A computer program product, comprising:
a non-transitory computer readable medium comprising code executable by a processor for performing operations comprising:
executing an application for a user, wherein executing an application for a user comprises:
determining, via a machine learning model, a priority of the application to the user, wherein the machine learning model determines the priority based, at least in part, on telemetry data, wherein the telemetry data describe user interactions performed with the application, tasks performed by the application, and resources utilized by the application;
subsequent to the determination of the priority of the application, placing the application on the whitelist;
based on the application being placed on the application whitelist allocating system resources of the information handling system to the application by performing steps comprising:
determining a macro classification for the application based on a resource utilization associated with the application from the application whitelist using a second model, wherein the second model is trained, at least in part, on the telemetry data and characteristic information associated with resource utilization for tasks performed by the application;
determining a system resources for functions being performed by the application using a third model, wherein the third model is trained, at least in part, on the telemetry data and real-time metrics associated with resources for tasks performed by the application;
allocating the system resources based, at least in part, on the macro classification and the system resources for tasks performed by the application;
updating usage of the allocated system resources of the information handling system by the application;
updating the macro classification for the application whitelist based, at least in part, on the updated usage of the system resources using the second model;
updating the resource utilization for tasks performed by the application based, at least in part, on the updated usage of the system resources using the third model; and
reallocating the system resources based, at least in part, on the updated macro classification and the updated resource utilization for tasks performed by the application.
8. The computer program product of claim 7 , wherein updating the application whitelist comprises updating the resource utilization associated with the application on the application whitelist.
9. The computer program product of claim 7 , wherein the third model is trained to predict resource usage by the application based on a function being performed by the application.