In recent years, Massively Parallel Processors have increasingly been used to manage and query vast amounts of data. Dramatic performance improvements are achieved through distributed execution of queries across many nodes. Query optimization for such system is a challenging and important problem.