A huge amount of real-life data has spatial attributes attached to it -- from addresses and phone numbers to specific location information like cellular phones' positions.
The importance of storing, accessing and analyzing location data increased dramatically with the emergence of the new mobile computing platforms and their associated applications. As a result, these spatial attributes are now exploited by an ever-increasing number of emerging applications, from traditional systems like Enterprise Resource Planning (ERP) systems to the new location-based services offered by wireless providers to enhance their offerings.
Allowing spatial data to be managed and accessed together with the regular non-spatial enterprise data is of crucial importance to the success of these systems, thus making the spatially extended relational database management system (RDBMS) a valuable container for location data.
The goal of the spatial database extensions project was to introduce new database technologies for storing, indexing, accessing and analysing abstract data types (ADT) and especially spatial data. These new technologies were implemented on HP’s NonStop SQL Regional Database Management Software (RDBMS) as a prototype demonstrating their advantages and strengths.
Problems Addressed
To efficiently manage spatial data in a RDBMS, the following issues should be addressed:
- Spatial indexing methods to efficiently support spatial query processing
- Efficient spatial index population and maintenance
- Spatial query optimization methods, that also support data partitioning and execution parallelism
- Representation of spatial data distribution and costing of spatial queries
|
 |
|
Spatial Extensions supports complicated queries that involve spatial joins between geometric features using various spatial functions.
|
|
Our Contribution
The main HP Labs' contribution in this project is the design of these new technologies and algorithms:
- New spatial indexing method, which can be implemented over the well established B-tree indexing, that efficiently supports Spatial query processing
- New scalable algorithms for spatial index population, and new fast algorithms for index maintenance
- New general query optimization methods for ADT and for spatial data in particular, that support scalable queries
- New spatial queries costing paradigm, and new database histograms to represent spatial data distribution.
In addition to this research, we implemented these new technologies and algorithms to extend the NonStop SQL database with spatial capabilities, as a prototype that demonstrates their efficiency and scalability properties.
Contact
Tal Drory
Emal: tal.drory@hp.com
Phone: +972-4-832-3050 x203
|
printable version
