Information processing system for scheduling jobs, job management apparatus for scheduling jobs, program for scheduling jobs, and method for scheduling jobs

A job management apparatus that manages job executed by one or more calculation nodes, the job management apparatus comprising:

• a storage device configured to store a first maximum power value which has been consumed when one or more calculation nodes has executed a first job;
• and
• a controller configured to detect the first job, which has been executed, whose identification information is matched with identification information of a second job to be scheduled, whose number of the calculation nodes to use is matched with a number of the calculation nodes of the second job, or whose difference of the number of the calculation nodes with the second job is within a prescribed range, predict, as a second maximum power value of the second job, the first maximum power value of the detected first job stored in the storage device, and schedule the second job such that the second maximum power value of the second job does not exceed a power consumption limit value set according to a time;
• and
• wherein the controller further detects, as the first job, a job whose type is matched with the second job besides the identification information.

A non-transitory computer-readable recording medium having stored therein a program for causing a computer to execute controlling process for a job management apparatus that manages jobs executed by one or more calculation nodes, the controlling process comprising:

• detecting the first job which has been executed, whose identification information is matched with an identification information of a second job to be scheduled, whose number of the calculation nodes to use is matched with a number of the calculation nodes of the second job, or whose difference of the number of the calculation nodes with the second job is within a prescribed range;
• predicting, as a second maximum power value of the second job, a first maximum power value which has been consumed when one or more calculation nodes has executed the detected first job;
• and
• scheduling the second job such that the second maximum power value of the second job does not exceed a power consumption limit value set according to a time;
• wherein the detecting further comprising second detecting, as the first job, a job whose type is matched with the second job besides the identification information.

A method of controlling an information processing system comprising a plurality of calculation nodes that executes jobs and a job management node that manages the jobs, the method comprising:

• detecting the first job which has been executed, whose identification information is matched with an identification information of a second job to be scheduled, of which the whose number of the calculation nodes to use is matched with a number of the calculation nodes of the second job, or whose difference of the number of the calculation nodes with the second job is within a prescribed range;
• predicting, as a second maximum power value of the second job, a first maximum power value which has been consumed when one or more calculation nodes has executed the detected first job;
• and
• scheduling the second job such that the second maximum power value of the second job does not exceed a power consumption limit value set according to a time;
• wherein the detecting further comprising second detecting, as the first job, whose type is matched with the second job besides the identification information.