Please Note: The master branch may be in an unstable state during development. Please use our releases for productive environments.

BBoxDB is a highly available distributed storage manager (NoSQL database), designed to handle multi-dimensional big data. Primarily, the software is a research project to explore new ways to handle multi-dimensional data in a distributed environment.

In contrast to existing key-value stores, BBoxDB can handle multi-dimensional efficiently. Existing key-value stores are using one-dimensional keys to address the values. Finding a proper key for multi-dimensional data is hard and often impossible; this is especially true when the data has an extent (non-point data / regions). To retrieve multi-dimensional data from a key-value store, a full data scan is often required. BBoxDB was developed to avoid the expensive full data scan and to make the work with multi-dimensional data more convenient.

• The well-known key-value data model is enhanced by a bounding box to describe the location of n-dimensional data in space.

• Data (point and non-point) of any dimension is supported.

• The data is indexed, which enables efficient range query processing.

• BigData is supported by spreading the data across a cluster of systems. Each node stores only a small part of the whole dataset.

• Multi-dimensional shards are created dynamically on the actual distribution of the data (automatically scale-up/scale-down).

• Data of multiple tables is stored co-partitioned, and spatial-joins can be executed efficiently without data shuffling between nodes.

• Data are re-distributed in the background without any service interruption.

• Continuous queries with bounding box query predicates are supported.

• User defined filters for query processing on custom data types.

The documentation of the project is located at https://jnidzwetzki.github.io/bboxdb/. The documentation also contains the changelog of the project.

For a guided tour through the features of BBoxDB, see the getting started chapter in the documentation. We also recommend reading the creating client code section. The install guide explains the needed steps to deploy an own BBoxDB cluster. The guide also describes how you can setup a virtualized cluster with 5 BBoxDB nodes in under two minutes, by using Docker and Docker Compose.

BBoxDB ships with a GUI that allows observing the global index structure. Below you find two screenshots of the GUI. The screenshots show how the space is partitioned. In addition, some details about the discovered nodes are shown.

(The screenshots contain content from OpenStreetMap - CC-BY-SA 2.0)

When 2-dimensional bounding boxes with WGS 84 coordinates are used, a map overlay visualization is supported by the GUI. On the top right picture, some spatial data about Germany was imported and the Figure shows, how Germany in partitioned after the data was imported. In addition, the GUI provides operations to explore 2 dimensional GeoJSON encoded data.

• Visit our website
• Read our research papers:
  • Technical report about the basic ideas of the software.
  • Demo paper about partitioning (presented at CIKM2018).
  • Full paper at Springer Distributed and Parallel Databases.
  • Demo paper about user defined filters (presented at IEEE Big Spatial Data 2019; received the best paper award).
• Join our chat at gitter
• Follow us on Twitter: @BBoxDB
• Subscribe our mailing list at Google Groups
• Visit our bug tracking system
• Read the source code and the documentation
• For contributing, see our contributing guide
• If you like the project, please star it on GitHub!

BBoxDB is licensed under the Apache 2.0 license. See the LICENSE file for details.