On-line analytical processing (OLAP) has become an increasingly important concern for telecommunications operators. Telecommunications systems generate huge amounts of data that would be beneficial to analyze using OLAP technology. Recently, Chatziantoniou et al. [4] presented the MD-join, a relational operator that provides a clean separation between group definitions and aggregate computations, and allows to succinctly express OLAP queries. In this paper, we define generalized MD-joins, describe an implementation of the GMD-join query engine on top of a commercial DBMS, and present a reduction of GMD-joins to SQL. We present a practical new optimization of GMD-joins allowing the restriction of base-values, and discuss the optimization of GMDjoins with restrictions and coalescing of GMD-joins. We show how GMD-join optimizations can be used to improve the performance of the translated SQL queries. Finally, we present experiments comparing the GMD-join query engine with the SQL reduction.