HADOOP PREEMPTIVE DEADLINE CONSTRAINT SCHEDULING METHOD, COMPUTER PROGRAM EXECUTING SAME, AND RECORDING MEDIUM IN WHICH PROGRAM IS RECORDED

摘要

The present invention relates to a Hadoop preemptive deadline constraint scheduling method, a computer program executing the same, and a recording medium in which the program is recorded. A scheduling method applied to a Hadoop system includes the steps of: (a) obtaining information about a plurality of jobs and performing initialization by scanning a queue, by a scheduler; (b) finding and scheduling an available slot by scanning the queue, by the scheduler; (c) determining whether to preempt a slot of a currently executed job by a new job, by calculating a remaining time to an execution completion time of the currently executed job to give the priority, and comparing the remaining time of the new job and a period until a deadline according to a sequence of the priority, by the scheduler; and (d) giving and allocating the slot to the job according to the determination of the preemption state, by the scheduler. The present invention as described above provides the Hadoop preemptive deadline constraint scheduling method capable of maximizing the number of the jobs performed under the deadline by effective use of the slots in a Hadoop environment, supporting the preemption, and improving preemption determination criteria for reducing an overhead of the preemption, thereby improving performance.;COPYRIGHT KIPO 2016