Existing highly precise density functional method for electronic structure calculations are mostly restricted to the treatment of at maximum a few hundred inequivalent atoms. This limitation leaves many open questions in material science e.g. on complex defects and defect-defect interaction unresolved. KKRnano is a new massively parallel DFT-algorithm in the framework of the KKR Green function method which we developed and optimized for large-scaled applications of thousands of atoms. In order to deal with the enormous computational requirements of such calculations we have implemented four levels of parallelization, which allows for an efficient use of hundreds of thousands of processors on the latest generation of supercomputers as Blue Gene. In this talk, I will present the advantages of the algorithm of KKRnano, the strategies for parallelization and optimization as well as the option to achieve linear scaling with system size.