Dynamic job scheduling system and method for supporting real-time stream data processing in distributed in-memory environment

摘要

According to the present invention, a dynamic job scheduling method in a dynamic job scheduling system for supporting real-time stream data processing in a distributed in-memory environment includes the steps of: calculating a load of each node in consideration of CPU usage, a size of an idle memory, and a network load when stream data is inputted; performing static job scheduling for assigning priority to each node according to the calculated load of each node to distribute jobs; managing status information of each node and load information consumed by each node in a load table after the distribution; and searching for a cause of a load by periodically checking the information of the load table, and redistributing the jobs through dynamic job scheduling using the CPU usage, memory usage, and the network load of each node according to the cause of the load. According to the present invention, a job delay and a loss due to a load on a node are prevented from occurring, and a data processing performance is improved.