GeoNames founder opens up the GIS world

What infrastructure do you use to run the site?

The site in particular the web services are now using three dedicated servers with Apache Tomcat and PostgreSQL running on Linux.

What search mechanism does the site use?

The GeoNames search is a custom search implementation based on the apache full text search library lucene. Lucene is a java search engine supporting inverted indices with term-frequency and inverse-document-frequency scoring. The GeoNames search implementation supports searches in all languages and it goes to great lengths to make sure the most relevant place is listed on top of the search engine result page.

Is Lucene responsible for making GeoNames so fast in retrieving, compiling and presenting data?

(Lucene is a) fantastic information retrieval library, but Postgres also has a lot of functions, operators and indexing possibilities for spatial data that make it blazingly fast.

What languages, coding, scripting and software do you and the development team use in working on GeoNames?

For the coding we mainly use java. The beauty of java is the tremendous number of fantastic open source libraries available for all different kind of tasks, like lucene for search. Furthermore we use Postgres/PostGIS with PROJ.4 for coordinates reprojections and the FWTools set bundled and developed by Frank Warmerdam.

What are the bandwidth and storage requirements of running the site?

At the moment the requirements are around 300GB traffic per month and 400GB storage.

How many active developers do you have?

There are about 30 GeoNames ambassadors, who help with questions regarding their countries and serve as local contact person for national data providers such as national mapping agencies, statistical offices or postal services. The pure development team consists of four or five people.

About how many downloads has the project had?

The files are downloaded about 15, 000 times per month. Considering that we had far fewer downloads in the beginning, I would estimate the total number of downloads to somewhere between 100, 000 and 200, 000

Will Autodesk's plans to donate its Mentor acquired coordinate system software to the open source community have any impact on GeoNames?

It might simplify things for GeoNames when reprojecting coordinates. I don't know yet how it compares to the open source PROJ.4 library. We don't really need exotic features. For us, these are tools we would like to have as user friendly as possible. We also see a trend towards WGS84 by data providers so that reprojecting coordinates will be less often needed in the future.