Ecosur will soon be providing students and researchers with a networked data base for storing and visualising public domain spatial layers. This database should also store results from ongoing research. I am convinced that the advantages of providing a fully functional geographical data base at zero cost by using PostGIS/Postgresql are compelling. The release of a stable Postgresql 8.3 and Qgis 0.10 have made the task even simpler.
Postgresql 8.3 is now notably faster when running spatial queries than the version I had been using (8.1). It also has very advanced features. These move PostgreSQL up to the power of the premium proprietary enterprise data base, Oracle. They clearly go beyond those found in MySQL.
The only practical challenge I faced with PostgreSQL 8.1 involved a very simple issue of scalability when the database was viewed with Qgis 0.9. Despite some issues with the stability of Qgis (that did worry me, but appear to have been largely resolved) Qgis is a suitable visualisation tool for PostGIS. I have used it extensively since I started experimenting with PostGIS a few months ago and am generally satisfied. However a simple practical issue concerned me.
How could users easily be given access to the parts of a potentially very large database in such a way that they see only the part they need or are allowed to see?
The answer had clearly to be through the use of Postgresql schemas, but the Qgis interface didn’t show them in a user friendly way that matched the need.
This problem is now largely resolved by the new version of Qgis. Schemas now show up in an unfoldable tree structure. As groups of users and individual users can be given access to a subset of schemas it should now be feasible to design a database that hides all the complexity from users and allows them to quickly find the layers they need.