Producing high quality recommendations has become a challenge in the recent years. Indeed, the growth in the quantity of data involved in the recommendation process pose some scalability and effectiveness problems. These issues have encouraged the research of new technologies. Instead of developing a new recommender system we improve an already existing method. A distributed framework was considered based on the known quality and simplicity of the MapReduce project. The Hadoop Open Source project played a fundamental role in this research. It undoubtedly encouraged and facilitated the construction of our application, supplying all tools needed. Our main goal in this research was to prove that building a distributed recommender system was not only possible, but simple and productive.