Federation, the ability to access diverse and distributed data as if it were a single source is a key requirement for information systems of today’s businesses. DB2 II, a highly efficient and configurable federated database system, provides outstanding performance with all major data sources. It performs so well because each federated source gives a precise model of its syntactical and semantical capabilities back to DB2. This information is then used to leverage individual strengths of both systems and to find the optimal way of distributing and executing heterogeneous queries. Without having that information, the optimizer would be ‘blind” and overall query performance would degrade by orders of magnitude. Unfortunately, this is exactly what happens if you federate data from an ODBC source. The reason lies in the nature of ODBC itself. It is an open standard that hides data source specifics from the client so that there is no possibility to return the model of data source capabilities to DB2 and let the optimizer leverage the strengths of both systems. This thesis proposes a way to automatically discover and tune federated engine parameters for ODBC datasources and proves, that this solution will significantly improve performance for virtually any ODBC source that is part of a federated system.