Back to results
US 12,585,501 B2Granted
Machine-learning (ML)-based resource utilization prediction and management engine
AssigneeBank of America Corporation
InventorsGopal Narayanan, Paparayudu Anaparti, Vishnu Vardhan Sarva, Sai Karthik Nanduri, Saikiran Gunti
Life of this patent
Estimating the term…
Abstract
Systems and methods for optimizing resource utilization in a computing system are provided. Methods include compiling training data, running a machine-learning (ML) model using the training data to determine a set of resource parameters, receiving a request to schedule the computing job on the computing system, computing, for each of a plurality of potential runtimes, an amalgam distance value, selecting the potential runtime with the optimal distance value for scheduling the computing job, and generating a job execution schedule based on the selected potential runtime.
Claims18
A machine-learning (ML)-based engine for optimizing resource utilization in a computing system, said engine comprising:
- a processor;
- a non-transitory memory storing computer executable instructions that, when run on the processor, are configured to:
- compile training data, the training data comprising historical resource utilization information;
- run an ML model using the training data to determine a set of resource parameters that exceed a predetermined threshold impact level in predicting future resource utilization by a computing job on the computing system;
- receive a request to schedule the computing job on the computing system;
- compute, for each of a plurality of potential runtimes, a distance value, wherein the distance value:
- represents a measurement of predicted resource utilization by the computing job on the computing system at the potential runtime;
- and
- is a single value that represents the measurement of predicted resource utilization across the set of resource parameters;
- automatically select the potential runtime with the optimal distance value for scheduling the computing job;
- and
- generate a job execution schedule based on the selected potential runtime;
- wherein the engine is further configured to:
- present the plurality of potential runtimes along with the distance values of the potential runtimes on a display of a computing device associated with a system administrator, wherein the display shows a 3-dimensional rendering of the distance values for each of the plurality of potential runtimes;
- receive a selection of a potential runtime from the system administrator via the computing device;
- and
- modify the job execution schedule based on the received selection.
14 more elements
A method for optimizing resource utilization in a computing system, the method comprising:
- compiling training data, the training data comprising historical resource utilization information;
- running a machine-learning (ML) model using the training data to determine a set of resource parameters that exceed a predetermined threshold impact level in predicting future resource utilization by a computing job on the computing system;
- receiving a request to schedule the computing job on the computing system;
- computing, for each of a plurality of potential runtimes, a distance value, wherein the distance value:
- represents a measurement of predicted resource utilization by the computing job on the computing system at the potential runtime;
- and
- is a single value that represents the measurement of predicted resource utilization across the set of resource parameters;
- automatically selecting the potential runtime with the optimal distance value for scheduling the computing job;
- and
- generating a job execution schedule based on the selected potential runtime;
- wherein the method further comprises:
- presenting the plurality of potential runtimes along with the distance values of the potential runtimes on a display of a computing device associated with a system administrator, wherein the display shows a 3-dimensional rendering of the distance values for each of the plurality of potential runtimes;
- receiving a selection of a potential runtime from the system administrator via the computing device;
- and
- modifying the job execution schedule based on the received selection.
12 more elements
A machine-learning (ML) system with optimized resource utilization, said system comprising:
- an ML model;
- a processor;
- a non-transitory memory storing computer executable instructions that, when run on the processor, are configured to:
- compile training data, the training data comprising historical resource utilization information;
- run the ML model using the training data to determine:
- a set of resource parameters that exceed a predetermined threshold impact level in predicting future resource utilization by a computing job on the system, wherein the set of resource parameters are selected from a list comprising execution time, memory, processing power, remote procedure call (RPC) connection, number of files, and data volume;
- and
- a weighting multiplier for each resource parameter in the set of resource parameters;
- receive a request to schedule the computing job on the system;
- compute, for each of a plurality of potential runtimes, a distance value, wherein the distance value:
- represents a measurement of predicted resource utilization by the computing job on the system at the potential runtime;
- and
- is a single value that represents the measurement of predicted resource utilization across the set of resource parameters, wherein the single value is computed using the weighting multipliers;
- automatically select the potential runtime with the optimal distance value for scheduling the computing job;
- and
- generate a job execution schedule based on the selected potential runtime;
- wherein the system is further configured to:
- present the plurality of potential runtimes along with the distance values of the potential runtimes on a display of a computing device associated with a system administrator, wherein the display shows a 3-dimensional rendering of the distance values for each of the plurality of potential runtimes;
- receive a selection of a potential runtime from the system administrator via the computing device;
- and
- modify the job execution schedule based on the received selection.
18 more elements
Dependencies are inferred from the claim text.