SpatialHadoop: A MapReduce Framework for Big Spatial Data
This talk describes SpatialHadoop; an open-source full-fledged system for indexing, querying, and visualizing big spatial data. SpatialHadoop is built as a comprehensive extension to Hadoop that injects spatial data awareness inside each Hadoop layer, namely, language, indexing, operations, and visualization. The language layer provides a simple high-level language with industry-standard spatial data types and functions. The indexing layer introduces a set of spatial indexes that can be built on big spatial datasets, such as, R-tree, Quad-tree, and K-d tree. The operations layer encapsulates a wide range of spatial operations including range query, spatial join, and computational geometry. The visualization layer provides an extensible visualization module that allows users to generate customized images to interactively explore big spatial datasets.
This talk will also describe three case studies of applications that use SpatialHadoop as a backbone to process big spatial data. SpatialHadoop is available for download at http://spatialhadoop.cs.umn.edu, along with setup instructions, tutorials, and real datasets to use.