Project Title: III: Small: Indexing, Querying, and Visualizing Big Spatial and Spatio-temporal Data Project Award Number: IIS-1525953 PI Name: Mohamed Mokbel Department: Computer Science and Engineering Institution: University of Minnesota Address: 200 Union ST SE, Minneapolis, MN, 55455, USA Email: mokbel@cs.umn.edu URL: www.cs.umn.edu/~mokbel |
|
|
||
  SpatialHadoop
is an open source MapReduce framework with built-in
support for spatial data. It employs the MapReduce programming paradigm
for distributed processing to build a general purpose tool for large
scale analysis of spatial data on large clusters. Users can interact
easily with SpatialHadoop through a high level language with built-in
support for spatial data types and spatial operations. Existing spatial
data sets can be loaded in SpatialHadoop with the built in spatial data
types point, polygon and rectangle. SpatialHadoop is also extensible
and more data types can be added by users. In addition, the data sets
are stored efficiently using built-in indexes (Grid file or R-tree)
which speed up the retrieval and processing of these data sets. Users
can build an index of their choice with a single command that runs in
parallel on the machines in the cluster. Once the index is built, users
can start analyzing their data sets using the built in spatial
operations (range query, k nearest neighbor and spatial join). The
extensibility of SpatialHadoop allows users to implement more spatial
operations as MapReduce programs. For more information, please visit: "http://spatialhadoop.cs.umn.edu/"
|
||
|
|
||
 
ST-Hadoop is a MapReduce framework that acknowledges the fact that space and time play a crucial role in query processing.
ST-Hadoop is an open-source extension of a Hadoop framework that injects the spatiotemporal awareness in the code base of four layers inside
SpatialHadoop, namely, language, indexing, MapReduce, and operations layers. The spatio-temporal indexing techniques inside ST-Hadoop primarily
tuned to provide the accommodation of new updated dataset efficiently without the need to rebuild its index.
The key point behind the performance gain of ST-Hadoop is the idea of indexing, where data are temporary loaded and divided across
computation nodes.
For more information, please visit: "http://st-hadoop.cs.umn.edu/"
|
||
