GT.M is a high end platform for developing transaction processing database applications. Individual database files in the terabyte range are supported; a logical database can consist of many files. It provides full support for ACID transactions using a Software Transaction Memory (STM) programming model and optimistic concurrency control. The GT.M data model is hierarchical associative memory (i.e. multi-dimensional arrays) that imposes no restriction on the data types of the indexes or content, nor any restriction on the schema is that imposed by the application. Such databases are also referred to as "document oriented", "schemaless", or "schema-free".
Ingres is a feature rich and robust database (RDBMS) that's free to use, modify, and redistribute. Support is available for a reasonable price if you need it. Ingres supports many programming languages, including Python, Ruby, Java, Perl, C/C++, and more. Ingres has a bit of a learning curve, but once mastered is extremely powerful and highly tunable. It is also very stable and reliable. Ingres is one of a fairly small number of databases providing excellent geospatial features including data types and functions compliant with the Open Geospatial Consortium (OGC) Simple Feature Specification (SFS) for SQL.