Method for predicting resource usage for applications in a distributed system
Inventors
Rowe, Anthony • Joe-Wong, Carlee • Pressler, Michael • PEREIRA, Nuno • Huang, Tianshu
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Assignees
Member
Carnegie Mellon UniversityCarnegie Mellon University is a global research institution based in Pittsburgh, Pennsylvania, recognized for interdisciplinary education, research, and innovation in science, engineering, arts, technology, and social sciences. The university leads advancements in artificial intelligence, robotics, digital health, and performing arts. Located in a technology-driven and culturally rich city, CMU powers real-world impact through research centers, industry engagement, workforce training, and initiatives that shape regional and global communities.
Carnegie Mellon University is a global research institution based in Pittsburgh, Pennsylvania, recognized for interdisciplinary education, research, and innovation in science, engineering, arts, technology, and social sciences. The university leads advancements in artificial intelligence, robotics, digital health, and performing arts. Located in a technology-driven and culturally rich city, CMU powers real-world impact through research centers, industry engagement, workforce training, and initiatives that shape regional and global communities.
Publication Number
US-12463919-B2
Publication Date
2025-11-04
Expiration Date
Abstract
A method for predicting resource usage for applications in a distributed system. The method includes: obtaining resource usage data, the resource usage data resulting from measuring the resource usage of different applications on different devices of the distributed system; detecting, by an orchestrator, a change and/or an event in the distributed system that requires a re-configuration of the distributed system; predicting, by the orchestrator, the resource usage of at least one application when deployed on one or different devices of the distributed system, the predicting being carried out based on the obtained resource usage data; initiating the required re-configuration based on the detecting and the predicting.
Core Innovation
An orchestrator predicts resource usage for applications in a distributed system. The orchestrator obtains resource usage data resulting from measuring resource usage of different applications on different devices of the distributed system, and based on the obtained resource usage data, predicts the resource usage of at least one application when deployed on one or different devices of the distributed system.
The orchestrator detects a change and/or an event in the distributed system that requires a re-configuration of the distributed system. The orchestrator initiates the required re-configuration based on the detected change/event and the predicted resource usage.
The prediction is carried out based on matrix factorization, using features for the applications and devices that are obtained and combined with features learned by the matrix factorization. The disclosed approach supports forming application and device features and combining them with features learned by matrix factorization for the predicting of resource usage.
Claims Coverage
The relevant independent claims are clm-00001, clm-00008, and clm-00009. Across these independent claims, the core claimed inventive elements include orchestrator-driven detection of a change/event that requires re-configuration, prediction of resource usage for an application on devices after deployment, and matrix-factorization-based prediction using combined application/device features and learned features.
Orchestrator-driven re-configuration loop using detected changes/events and predicted resource usage
An orchestrator obtains resource usage data from measuring resource usage of different applications on different devices, detects a change and/or an event in the distributed system that requires a re-configuration, predicts resource usage of at least one application when deployed on one or different devices based on the obtained data, and initiates the required re-configuration based on the detecting and the predicting.
Matrix-factorization-based resource usage prediction using combined application and device features with learned features
The resource usage predicting is carried out based on matrix factorization, wherein features for the applications and devices are obtained and combined with features learned by the matrix factorization for the predicting of the resource usage.
Non-transitory computer program for performing orchestrator-driven matrix-factorization prediction and re-configuration
A non-transitory computer-readable medium stores a computer program including instructions that, when executed, perform obtaining measured resource usage data, orchestrator detection of a change/event requiring re-configuration, orchestrator prediction of resource usage based on matrix factorization with combined application/device features and learned features, and initiation of the required re-configuration based on the detecting and predicting.
Data processing apparatus configured to obtain resource usage data, detect reconfiguration-triggering events, and predict via matrix factorization
A data processing apparatus is configured to obtain resource usage data from measuring resource usage of different applications on different devices, detect by an orchestrator a change/event requiring re-configuration, predict by the orchestrator resource usage of at least one application when deployed on one or different devices based on the obtained data, and initiate the required re-configuration based on the detecting and predicting, wherein the predicting is carried out based on matrix factorization using obtained and combined application/device features with features learned by the matrix factorization.
The independent claims collectively cover an orchestrator-based mechanism that detects changes/events requiring re-configuration and predicts application resource usage on devices. The prediction is specifically performed using matrix factorization that combines application/device features with features learned by the matrix factorization, and the output prediction is used together with the detected change/event to initiate the required re-configuration.
Stated Advantages
Not explicitly described in patent.
Documented Applications
Not explicitly described in patent.
Interested in licensing this patent?