Issue link: https://read.dmtmag.com/i/35322
Do NOSQL and Geospatial Technologies Have a Future Together? BUILDINGTHEGEOWEB O ne of the more unexpected developments from the social-media revolution has been the rise to prominence of NOSQL data stores such as Dynamo, Hbase and Cassandra. Although the interest in non-relational data stores certainly didn’t begin with social media, they have proven their worth and gathered the greatest amount of attention in this domain. BY RON LAKE What’s NOSQL? As with any new technology, the term NOSQL is imprecise, and there’s no widely accepted definition. In general, a NOSQL data store is a database that’s not fundamentally built on relations, includ- ing Column-Oriented Data Stores, DOM-Oriented Native XML databases and Key-Value Data Stores— a wide gamut indeed. With the long history and evident success of rela- tional technology (e.g., Oracle, MS, IBM, MySQL), why would non-relational databases need to be developed? The impedance mismatch of objects and relations debated in the 1990s didn’t stop relational technology, nor did the development of query languages superior to SQL, such as OQL and, more recently, XQuery. Some of the relational dominance can be put down to the sheer market size that had been acquired by relational software companies by the time these early contenders appeared (e.g., object databases in the late 1980s). The big guys were able to “roll with the punches” and absorb their competi- tors (e.g., IBM eats Informix eats Illustra, etc.) or extend their own products (e.g., Oracle 8i). Ron Lake is president, Galdos Systems Inc.; e-mail: rlake@ galdosinc.com, blog: www. galdosinc.com/ archives/category/ media-center/blog. 30 Here to Stay This time, things look a little different. Not that the big database companies are going anywhere, but it seems that NOSQL data stores will indeed gain a foothold in the marketplace. Unlike their predecessors, they have proven their scalability and utility in very large installa- tions (multi-terabytes, petabytes and beyond). Given such furor, and some significant successes, NOSQL data stores likely will have implications for geospatial databases, especially in applications GEO W ORLD / JUL Y 2O11 involving very large datasets and where data schemas (relational DBMSs) often are changing. Several geospatial applications seem promising, including the hosting (clouds?) of large 3-D city models, large image and LIDAR data stores, detailed epidemiology databases, and local or global environ- mental models. These applications certainly require terabytes or more of storage. Large and detailed city models will be the new foundation for real and virtual views of our world. Many NOSQL databases have shown their ability to scale horizontally, with systems comprising tens of thousands of commodity servers. This also is of great interest in the geospatial world, and not just for Google Earth. Think of having detailed models of every city in a country, with sufficient captured semantics to drive everything from urban design to augmented reality to climate-change impact analysis. Urban Modeling Applications All these developments have, of course, significant resonance for CityGML and IFC, the two primary seman- tic models for modeling urban environments. A single city model may incorporate models of hundreds of thou- sands of structures. The Berlin city model (CityGML), for example, has modeled more than 400,000 build- ings. Although this currently is only at Level of Detail 2 (LOD2), other jurisdictions are pushing the detail envelope to many thousands of buildings at LOD4. If this is combined with real-time sensor data, data volumes for a single urban center can easily push into the many-terabyte range. As more urban- modeling applications are developed, more things become possible, driving larger and more-detailed models that, in turn, drive more applications. Deploying and maintaining large urban models will require high-performance data stores, with automated content maintenance combined with fine-grained publishing to a distributed group of stakeholders worldwide (think of construction, engi- neering and development companies, not to mention augmented-reality consumers and game players). Such large and detailed city models will be the new foundation for real and virtual views of our world. The dream of a 1:1-scale digital model of the world, updated in real time, still is far in the future. Or is it? NOSQL data stores will have a role to play as technol- ogy moves ever more rapidly in that direction.