A parallel distributed data processing method for massive data visualization by data locality, and system thereof

摘要

The present invention relates to a method and a system for location-based distributed data processing to visualize massive data, which are capable of minimizing network costs required to process data visualization by processing data stored in a local memory first using the distribution information of data when visualizing massive data, and bringing and processing data shared by other nodes in an order in which the size information of the data increases. The system for location-based distributed data processing to visualize massive data according to the present invention includes: a distributed shared memory which retains a data block list for each node regarding data blocks required for visualization, and stores data blocks shared by each node; and a plurality of nodes which includes a local memory for loading the data blocks, stores data blocks stored in a disc in the local memory or the distributed shared memory and manages the data blocks, renders the data blocks stored in the local memory first, based on the data block list for each node, according to distributed jobs, brings, from the distributed shared memory, the shared data blocks in an order in which the size information of the data blocks retained by other nodes increases, and renders the shared data blocks in parallel.