Back to results
US 11,693,668 B2Granted
Parallel processing apparatus, storage medium, and job management method
AssigneeFujitsu Limited
InventorAkitaka Iwata
Life of this patent
Estimating the term…
Abstract
A parallel processing apparatus includes a plurality of compute nodes, and a job management device that allocates computational resources of the plurality of compute nodes to jobs, the job management device including circuitry configured to determine a resource search time range based on respective scheduled execution time periods of a plurality of jobs including a job being executed and a job waiting for execution, and search for free computational resources to be allocated to a job waiting for execution that is a processing target among the plurality of jobs, from among computational resources of the plurality of compute nodes within the resource search time range, by backfill scheduling.
Claims15
A parallel processing apparatus comprising:
- a plurality of compute nodes;
- and
- a job management device that allocates computational resources of the plurality of compute nodes to jobs, the job management device including circuitry configured to:
- calculate a resource search time range based on the largest value from among respective scheduled execution time periods of a plurality of jobs, each of the plurality of jobs being a job existing at a time when the calculating of the resource search time range is performed, each of the plurality of jobs being a job being executed or a job waiting for execution;
- search for, within the calculated resource search time range, first free computational resources from among computational resources of the plurality of compute nodes, to determine whether the first free computational resources is allowed to be allocated to a job waiting for execution that is a processing target among the plurality of jobs;
- in a case where the first free computational resources is allowed to be allocated to the job, allocate the first free computational resources to the job;
- in a case where the first free computational resources is not allowed to be allocated to the job;
- calculate a backfilling permission time range being a time range from the start time point of the resource search time range to a time point at which all of one or more jobs being executed end;
- and
- search for, by backfill scheduling within the calculated backfilling permission time range, second free computational resources from among computational resources of the plurality of compute nodes, in order to allocate the second free computational resources to the job waiting for execution that is the processing target.
7 more elements
A non-transitory computer-readable storage medium storing a job management program for causing a computer to execute a process, the process comprising:
- in allocating computational resources of a plurality of compute nodes to a job, calculating a resource search time range based on the largest value from among respective scheduled execution time periods of a plurality of jobs, each of the plurality of jobs being a job existing at a time when the calculating of the resource search time range is performed, each of the plurality of jobs being a job being executed or a job waiting for execution;
- searching for, within the calculated resource search time range, first free computational resources from among computational resources of the plurality of compute nodes, to determine whether the first free computational resources is allowed to be allocated to a job waiting for execution that is a processing target among the plurality of jobs;
- in a case where the first free computational resources is allowed to be allocated to the job, allocating the first free computational resources to the job;
- and
- in a case where the first free computational resources is not allowed to be allocated to the job;
- calculating a backfilling permission time range being a time range from the start time point of the resource search time range to a time point at which all of one or more jobs being executed end;
- and
- searching for, by backfill scheduling within the calculated backfilling permission time range, second free computational resources from among computational resources of the plurality of compute nodes, in order to allocate the second free computational resources to the job waiting for execution that is the processing target.
5 more elements
A job management method executed by a computer, comprising:
- by the computer,
- in allocating computational resources of a plurality of compute nodes to a job, calculating a resource search time range based on the largest value from among respective scheduled execution time periods of a plurality of jobs, each of the plurality of jobs being a job existing at a time when the calculating of the resource search time range is performed, each of the plurality of jobs being a job being executed or a job waiting for execution;
- searching for, within the calculated resource search time range, first free computational resources from among computational resources of the plurality of compute nodes, to determine whether the first free computational resources is allowed to be allocated to a job waiting for execution that is a processing target among the plurality of jobs;
- in a case where the first free computational resources is allowed to be allocated to the job, allocating the first free computational resources to the job;
- and
- in a case where the first free computational resources is not allowed to be allocated to the job;
- calculating a backfilling permission time range being a time range from the start time point of the resource search time range to a time point at which all of one or more jobs being executed end;
- and
- searching for, by backfill scheduling within the calculated backfilling permission time range, second free computational resources from among computational resources of the plurality of compute nodes, in order to allocate the second free computational resources to the job waiting for execution that is the processing target.
6 more elements
Dependencies are inferred from the claim text.