US 12,730,681 B2Granted

Efficient assignment of resource-intensive computing jobs

AssigneeSap SE

InventorsWei Jiang, Hao Weng

Life of this patent

Estimating the term…

Loading patent…

Abstract

Systems and methods described herein relate to the handling of resource-intensive computing jobs in a cloud-based job execution environment. An unexecuted computing job has a plurality of features. A resource intensity prediction is generated for the unexecuted computing job based on the features and on historical job data that classifies each of a plurality of executed computing jobs as either resource intensive or non-resource intensive. The resource intensity prediction indicates that the unexecuted computing job is predicted to be classified as resource intensive. A predicted resource intensity category of the unexecuted computing job is determined. Utilization data associated with one or more of a plurality of job execution destinations may be accessed. The unexecuted computing job may be assigned to a selected job execution destination from among the plurality of job execution destinations based on the predicted resource intensity category and the utilization data.

Claims17

Dependent claims

A system comprising:

at least one memory that stores instructions;
and
one or more processors configured by the instructions to perform operations comprising:
generating, based on historical job data and a plurality of features of an unexecuted computing job, a resource intensity prediction indicating that the unexecuted computing job is predicted to be classified as resource intensive, the historical job data classifying each of a plurality of executed computing jobs as either resource intensive or non-resource intensive;
determining a predicted resource intensity category of the unexecuted computing job, the predicted resource intensity category being selected from a plurality of resource intensity categories;
accessing utilization data associated with one or more of a plurality of cloud-based job execution destinations;
and
assigning, based on the predicted resource intensity category and the utilization data, the unexecuted computing job to a selected job execution destination from among the plurality of cloud-based job execution destinations, wherein the assigning of the unexecuted computing job to the selected job execution destination comprises:
identifying an assignment rule corresponding to the predicted resource intensity category, wherein at least some of the resource intensity categories have different assignment rules;
and
applying the assignment rule corresponding to the predicted resource intensity category to select the selected job execution destination from among the plurality of cloud-based job execution destinations.

8 more elements

A method comprising:

generating, based on historical job data and a plurality of features of an unexecuted computing job, a resource intensity prediction indicating that the unexecuted computing job is predicted to be classified as resource intensive, the historical job data classifying each of a plurality of executed computing jobs as either resource intensive or non-resource intensive;
determining a predicted resource intensity category of the unexecuted computing job, the predicted resource intensity category being selected from a plurality of resource intensity categories;
accessing utilization data associated with one or more of a plurality of cloud-based job execution destinations;
and
assigning, based on the predicted resource intensity category and the utilization data, the unexecuted computing job to a selected job execution destination from among the plurality of cloud-based job execution destinations, wherein the assigning of the unexecuted computing job to the selected job execution destination comprises:
identifying an assignment rule corresponding to the predicted resource intensity category, wherein at least some of the resource intensity categories have different assignment rules;
and
applying the assignment rule corresponding to the predicted resource intensity category to select the selected job execution destination from among the plurality of cloud-based job execution destinations.

5 more elements

A non-transitory computer-readable medium that stores instructions that, when executed by one or more processors, cause the one or more processors to perform operations comprising:

generating, based on historical job data and a plurality of features of an unexecuted computing job, a resource intensity prediction indicating that the unexecuted computing job is predicted to be classified as resource intensive, the historical job data classifying each of a plurality of executed computing jobs as either resource intensive or non-resource intensive;
determining a predicted resource intensity category of the unexecuted computing job, the predicted resource intensity category being selected from a plurality of resource intensity categories;
accessing utilization data associated with one or more of a plurality of cloud-based job execution destinations;
and
assigning, based on the predicted resource intensity category and the utilization data, the unexecuted computing job to a selected job execution destination from among the plurality of cloud-based job execution destinations, wherein the assigning of the unexecuted computing job to the selected job execution destination comprises:
identifying an assignment rule corresponding to the predicted resource intensity category, wherein at least some of the resource intensity categories have different assignment rules;
and
applying the assignment rule corresponding to the predicted resource intensity category to select the selected job execution destination from among the plurality of cloud-based job execution destinations.

5 more elements

Dependencies are inferred from the claim text.